A grammar of the Kuku Yalanji language of north Queensland - Elisabeth Patz.md

A grammar of the Kuku Yalanji language of north Queensland

Also in Pacific Linguistics

• Mark Harvey, 2001, A grammar of Limilngan: a language of the Mary River region, Northern Territory.
• Jane Simpson, David Nash, Mary Laughren, Peter Austin and Barry Alpher, 2001, Forty years on: Ken Hale and Australian languages.

David Rose, 2001, The Western Desert code: an Australian cryptogrammar.

L.A. Hercus, 1999, A grammar of the Wirangu language from the West Coast of South Australia.

Pacific Linguistics is a publisher specialising in grammars and linguistic descriptions, dictionaries and other materials on languages of the Pacific, the Philippines, Indonesia, Southeast and South Asia, and Australia.

Pacific Linguistics, established in 1963 through an initial grant from the Hunter Douglas Fund, is associated with the Research School of Pacific and Asian Studies at The Australian National University. The Editorial Board of Pacific Linguistics is made up of the academic staff of the School's Department of Linguistics. The authors and editors of Pacific Linguistics publications are drawn from a wide range of institutions around the world. Publications are refereed by scholars with relevant expertise who are usually not members of the editorial board.

The 470 publications before Publication 501 were numbered in four series, A, B, C and D, but the series were merged from the beginning of the year 2000 because the rationale for them had gradually disappeared.

FOUNDING EDITOR:	Stephen A. Wurm
EDITORIAL BOARD:	John Bowden, Malcolm D. Ross and Darrell T. Tryon (Managing Editors), I Wayan Arka, Andrew Pawley, Paul Sidwell

Pacific Linguistics 527

Heather Bowe and Stephen Morey, 1999, The Yorta Yorta (Bangerang) language of the Murray Goulburn including Yabula Yabula.

^Agrammar of the Kuku Yalanji language of north Queensland

Elisabeth Patz

Pacific Linguistics Research School of Pacific and Asian Studies The Australian National University

Published by Pacific Linguistics Research School of Pacific and Asian Studies The Australian National University Canberra ACT 0200 Australia

Copyright © The author

First published 2002

National Library of Australia Cataloguing-in-Publication entry:

Elisabeth Patz A grammar of the Kuku Yalanji language of north Queensland

Bibliography. ISBN 0 85883 534 7

1. Kuku-Yalanji language - Grammar. 2. Queensland, Northern - Languages. I. The Australian National University. Research School of Pacific and Asian Studies. II. Title. (pacific Linguistics 526).

499.15

Copyedited by Basil Wilson Typeset by Jeanette Coombes Cover design by Emily Brissenden Printed and bound by Union Offset Printers, Fyshwick, Canberra

For Ross Thank you

Table of contents

	List of maps, figures and tables	xiii
	Abbreviations and conventions	xv
	Acknowledgements	xviii
	Foreword	xix
	Chapter 1: Introduction	1
1.1	The language and its speakers	1
	1.1.1 Kuku Yalanji and surrounding languages	1
	1.1.2 Kuku Yalanji and its dialects	1
	1.1.3 Territory	6
	1.1.4 Present speakers	6
	1.1.5 Studies of Kuku Yalanji	8
1.2	The culture of the Kuku Yalanji	10
	1.2.1 Cultural background	10
	1.2.1.1 Tribal groups and names	11
	1.2.1.2 Personal names	11
	1.2.1.3 Social structure and marriage	11
	1.2.1.4 Food	12
	1.2.1.5 Artefacts	13
	1.2.1.6 Medicine	13
	1.2.1.7 Taboos	13
	1.2.1.8 Other customs and beliefs	14
	1.2.2 Recent history	15
	1.2.3 Summary	17
	Chapter 2: Phonology	18
2.1	Phoneme inventory	18
	2.1.1 Consonants	18
	2.1.2 Vowels	21
2.2	Phonotactics	22

viii

2.2.1	Possibility of occurrence	22
2.2.2	Probability of occurrence	25
2.3	Phonological adaptation of English loan words	28
2.4	Phonetic rules – assimilation	30
2.5	Morphophonological processes	31
2.5.1	Vowel harmony	31
2.5.2	Deletion of semivowel y	33
2.5.3	Syllable reduction	34
2.6	Stress and intonation	35
2.6.1	Word stress	35
2.6.2	Clause stress	36
2.6.3	Intonation contours	40
	Chapter 3: Morphology
3.1	Parts of speech	42
3.1.1	Formal and grammatical criteria	42
3.1.2	Semantic content	43
3.2	Nominal morphology	44
3.2.1	Case inflections – system	45
3.2.2	Case inflections – form	45
3.2.2.1	Historical notes	50
3.2.3	Nominal stem formation	50
3.2.3.1	Catalytic suffix -mun-	51
3.2.3.2	Possessive	51
3.2.3.3	Comitative	52
3.2.3.4	Privative	54
3.2.3.5	Number	55
3.2.3.6	Nominalising suffix -nyV	57
3.2.3.7	Other noun derivations	57
3.2.3.8	'Degree' of adjective	60
3.2.3.9	Co-occurrence of stem-forming affixes	61
3.2.3.10	Nominal compounds	63
3.2.4	Morphology of kinship terms	64
3.2.5	Morphology of quantifiers	65
3.3	Morphology of location words	67
3.3.1	Locational systems	68
3.3.2	Location words – inflections	69
3.4	Morphology of time words	70
3.5	Personal pronouns	72
3.5.1	Paradigm	72
3.5.2	Possessive stems	73
3.5.3	Historical notes	77
3.6	Interrogative pronouns	78
3.6.1	Paradigm and morphology	78
3.6.2	Historical notes	82
3.7	Demonstratives	83
3.8	Morphology of verbs	86
3.8.1	Introduction	86
3.8.2	Verbal word structure	86
3.8.2.1	General	86
3.8.2.2	Trisyllabic roots	87
3.8.3	Transitivity and conjugations	88
3.8.3.1	Correlation of transitivity and conjugation	88
3.8.3.2	Development of Kuku Yalanji conjugations	89
3.8.4	Verbal inflections	92
3.8.4.1	Paradigm	92
3.8.4.2	Functions of verbal inflections	92
3.8.5	Verbal stem formation	97
3.8.5.1	Causative compounds	97
3.8.5.2	Idiomatic verbal compounds	100
3.8.5.3	Inchoative stems	101
3.8.5.4	Syntactic derivations	103
3.8.5.5	Non-syntactic derivations	105
3.8.6	Adverbial forms	108
3.8.6.1	Adverbial particles	108
3.8.6.2	Derived adverbs	110
3.9	Free-form particles	111
3.9.1	Adjectival particles	111
3.9.2	Comparison, negation and discourse particles	112
3.10	Clitics	114
3.10.1	Temporal clitics	114
3.10.2	Emphatic clitics	115
3.10.3	Other clitics	116
3.10.4	Ordering of clitics	117
3.11	Interjections	117
3.12	Morphological adaptation of English loan words	117
Chapter 4:	Syntax	119
4.1	Simple sentences	119
4.1.1	Noun phrases	119
4.1.2	The verb complex	121
4.1.3	Verbs - syntactic types	121
4.1.4	Case functions	122
	4.1.4.1 Preliminary remarks	122
	4.1.4.2 Semantic functions of case inflections	123
	4.1.4.3 Core syntactic cases	127
	4.1.4.4 Peripheral syntactic and local cases	129
4.1.5	Nominal sentences	137
4.1.6	Order of sentence constituents	139
4.2	Clause-level derivations	140
	4.2.1 Inchoative constructions	140
	4.2.2 Causative constructions	141
	4.2.3 Reciprocal	143
	4.2.4 Transitive verb + $-ji-$ constructions	144
	4.2.4.1 Preliminary remarks	144
	4.2.4.2 Reflexive	145
	4.2.4.3 Passive	147
	4.2.4.4 Antipassive	151
4.2.5	Summary and conclusions	155
4.3	Sentence co-ordination	156
	4.3.1 Simple conjoining	156
	4.3.2 Co-ordination particles	158
	4.3.3 Clitics in co-ordinate constructions	160
4.4	Subordinate clauses	161
	4.4.1 Purposive	161
	4.4.2 Precautionary	167
	4.4.3 Successive	169
	4.4.4 Verb + $-nyV$ constructions	171
	4.4.4.1 Simultaneous action	171
	4.4.4.2 Causal	176
	4.4.4.3 Prior event	178
	4.4.4.4 Hypothetical reason	179
	4.4.5 Relative clauses	181
	4.4.6 Summary	183
4.5	Possession	185
	4.5.1 General (alienable) possession	185
	4.5.2 Inalienable possession	186
	4.5.3 Inanimate possessor	187
	4.5.4 Former possession	187
	4.5.5 Relationship between possessive and locative	188
4.6	Comitative and privative	189
	4.6.1 Comitative	189
	4.6.2 Privative	192

1X

x

4.7	Imperatives	194
4.8	Questions	195
4.9	Particles	196
	4.9.1 Adjectival particles	196
	4.9.2 Negation particles	197
	4.9.3 Other particles	199
4.10	Interjections	200
Chapter 5: Narrative discourse
5.1	Preliminary remarks	201
5.2	Cohesion	201
	5.2.1 Participant identification	202
	5.2.1.1 Establishing participant reference	202
	5.2.1.2 Maintaining participant reference	204
	5.2.1.3 Summary	206
	5.2.2 Temporal cohesion	206
	5.2.3 Anaphoric linkage	208
5.3	Topics	208
	5.3.1 Preliminary remarks	208
	5.3.2 Topic choice and topic marking	209
	5.3.2.1 Word order and syntactic function	209
	5.3.2.2 Switch function	210
	5.3.2.3 Choice of referent as topic	211
	5.3.2.4 Conclusions	212
5.4	Localised highlighting	213
	5.4.1 Specifically marked word order	213
	5.4.2 Emphatic clitics	214
5.5	Events and non-events	215
	5.5.1 Basic distinction between events and non-events	216
	5.5.1.1 Important events in Yalanji dialect	216
	5.5.1.2 Non-events versus events in Nyungkul dialect	216
	5.5.2 Spatial and temporal setting	217
	5.5.3 Collateral: telling what did not happen	213
5.6	Concluding remarks	220
Appendix: Texts
	Text 51: Nyungkul
	Charlie Tayley: Why the dingo always turns around before running away
	Text 12: Nyungkul
	Charlie Tayley: A story from the time of the Palmer River gold rush
Text 36: Nyungkul
Bobby Roberts: A prophetic dream	235
Text 3: Yalanji
Ivy Walker: A trip to Shipton's Flat (excerpts)	240
References	246

Xl

.--------------------_._----------------------------------------,

List of maps, figures and tables

Maps
1	Kuku Yalanji and its neighbours	2
2	The dialects of Kuku Yalanji	4
3	Nyungkul and Yalanji territory	7
Figures
2.1	Range of phonetic realisation of Kuku Yalanji vowels	22
2.2	Initial and final consonants in Kuku Yalanji	23
Tables
1.1	Attested dialects of Kuku Yalanji	5
2.1	Kuku Yalanji consonants	19
2.2	Heterorganic consonant clusters in Kuku Yalanji	24
2.3	Relative frequency of Kuku Yalanji vowels	26
2.4	Actual-to-expected ratio of vowel co-occurrence in first and second vowel position	26
2.5	Relative association between vowels in Kuku Yalanji	26
2.6	Frequency rates of initial and final consonants by place and manner of articulation	27
2.7	Vowel harmony in Wagaman inflections	32
3.1	Kuku Yalanji case inflections	47
3.2	Possessive suffixes in Yalanji and Nyungkul	53
3.3	Allomorphs of comitative suffix in Kuku Yalanji	53
3.4	Co-occurrence of nominal derivational suffixes in Kuku Yalanji	63
3.5	Case inflections and derivational suffixes with cardinal numbers	67
3.6	Personal pronoun paradigm	74
3.7	Personal pronoun base forms and inflections	75
3.8	Possessive pronoun stems with non-zero inflection	76
3.9	Interrogative pronouns in Yalanji and Nyungkul	78

	xiv

3.10	Paradigm of interrogative pronouns 'who' and 'what'	79
3.11	The forms of wanja 'where'	81
3.12	Demonstratives in Yalanji and Nyungkul	83
3.13	Paradigm of demonstrative pronouns	84
3.14	Paradigm of local demonstratives	85
3.15	Relationship between conjugation, transitivity, final root vowel, and number of syllables	89
3.16	Paradigm of verbal inflections	92
3.17	Inflected verbs	92
4.1	Animacy hierarchy and case marking	124
4.2	Case marking on noun phrase constituents other than head nouns	125
4.3	Functions and forms of subordinate clauses in Kuku Yalanji	184
5.1	Syntactic functions of topic in introduction and subsequent clauses with zero anaphora	211

Abbreviations and conventions

Examples consist of three parts:

(a) Kuku Yalanji sentence

The following conventions are used:

• morpheme boundary
• / alternative forms
• ( ) optional material
• { } different morphemes in Yalanji and Nyungkul dialects
• " " direct speech
• clause boundary in co-ordinate and subordinate constructions
• * ungrammatical sentence or word form
• ? not attested but presumably possible sentence, or construction that is considered awkward although grammatical
• primary and secondary word stress respectively
• I " primary and secondary clause stress respectively

(b) Interlinear glosses

These are lined up with the beginning of each Kuku Yalanji word. As far as possible direct translations or semantic approximations have been used for free and bound morphemes alike. Where such translations would have resulted in confusion with other lexical items, abbreviations of the grammatical function have been used. Groups of bound morphemes with similar function but different semantic contents are also indicated by abbreviations referring to the respective function. A list of abbreviations is set out below.

(c) Translations

The following conventions are used:

• items that can be inferred from the Kuku Yalanji sentence
• ( ) editorial comments
• / alternative translations
• " " verbatim translations by speakers or, in context, direct speech

For stylistic reasons English translations contain article the, regardless of whether the Kuku Yalanji example implies definiteness or not. For ease of reading the is not enclosed in parentheses.

Abbreviations are given below in alphabetical order together with the number of the section in which they are introduced:

(A)	transitive subject, Agent (function)	4.1.4.1
ABESS	abessive (case)	3.2.1
ABL	ablative (case)	3.2.1
ABS	absolutive (case)	3.2.1
ACC	accusative (case)	3.5.1
ADV	adverbalising suffix -ku	3.8.6.2
CAUS	causative (forming transitive compound verb from a nominal or an intransitive verb)	3.8.5.1
COMIT	comitative (nominal derivation)	3.2.3.3
COMP	comparison particle yala	3.9.2
DAT	dative (case)	3.2.1
DEG1	adjectival degree prefix jarra-	3.2.3.8
DEG2	adjectival degree suffix -baja	3.2.3.8
EMPH1	emphatic clitic -ku	3.10.2
EMPH2	emphatic clitic -lu, -la	3.10.2
EMPH3	emphatic clitic -bi	3.10.2
EMPH4	emphatic clitic -(V)rrku	3.10.2
ERG	ergative (case)	3.2.1
IMP	imperative (verb inflection)	3.8.4.1
INCHO	inchoative (forming intransitive verb stems or compounds from nominals)	3.8.5.3
INST	instrumental (case)	3.2.1
INTER	interjection	3.11
IRR	irrealis (verb inflection)	3.8.4.1
ITR, itr	intransitive (syntactic verb derivation)	3.8.5.4
KPL	kinship plural (nominal stem forming suffix)	3.2.3.5
LOC	locative (case)	3.2.1
mun	catalytic nominal suffix -mun-	3.2.3.1
NEG	negation particle kari	3.9.2
NEG2	negation particle balu	3.9.2
NOM	nominative (case)	3.5.1
NONPAST	nonpast tense (verb inflection)	3.8.4.1
(0)	transitive object (function)	4.1.4.1
PAST	past tense (verb inflection)	3.8.4.1
PERL	perlative (case)	3.2.1
PLS	plural subject agreement (verb suffix)	3.8.5.5
POSS	possessive (nominal derivation)	3.2.3.2
PRECAUT	precautionary (verb inflection)	3.8.4.1
PRIV	privative (nominal derivation)	3.2.3.4
:pt	'potent' case inflection	4.1.4.2
PURP	purposive (verb inflection)	3.8.4.1
RECIP	reciprocal (syntactic verb derivation)	3.8.5.4
REDUP	reduplication of nominals and verbs	3.2.3.5 and 3.8.5.5
REF	reflexive (syntactic verb derivation)	3.8.5.4
(S)	intransitive subject (function)	4.1.4.1
SUB	subordinate/nominalising verb suffix	3.2.3.6 and 3.8.4.1
SUCC	successive (verb inflection)	3.8.4.1
TEMP1	temporal clitic -da	3.10.1
TEMP2	temporal clitic -kuda	3.10.1
TEMP3	temporal clitic -(ng)Vrr	3.10.1
TEMP4	temporal clitic wawu-	3.10.1
UM	unmarked (verb inflection)	3.8.4.1

Abbreviations of pronouns

• 1 first person
• 2 second person
• 3 third person
• sg singular
• du dual
• pI plural
• inc inclusive
• exc exclusive

Abbreviations of kinship terms

• eB elder brother
• eZ elder sister
• FF father's father
• FM father's mother
• FZ father's sister
• MB mother's brother
• MF mother's father
• MiL mother-in-law
• MM mother's mother
• yB younger brother
• yz younger sister

Acknowledgements

I wish to record my deepest gratitude to the Kuku Yalanji people at Wujal Wujal, who accepted me with friendliness and generosity both as a researcher and as a temporary resident in their community. I cannot name all of my helpers here, but I owe special thanks to Charlie Tayley, Bobby Roberts, Louisa Smith, Ruby Friday, Ivy Walker and my babarr Mabel Webb for their unflagging enthusiasm in teaching me their language with intelligence, sensitivity and patience. Long hours of repetitive and tiring work were alleviated by their sense of humour, entertaining anecdotes and activities like forays for bush tucker in which they gladly allowed me to participate. Kuku Yalanji speakers of the Mossman Gorge Mission and the Mareeba settlement, in particular Norman Mitchell, also contributed material.

Hank and Ruth Hershberger of the Summer Institute of Linguistics kindly co-operated with me during fieldwork, comparing notes and making some of their unpublished material a vailable to me.

At The Australian National University I was very fortunate to have the assistance of Professor R.M.W. Dixon, who had introduced me to Australian linguistics, and Dr Harold Koch. They supervised my study with interest and enthusiasm, providing constructive comments and guidance.

During fieldwork I received financial and material assistance in the form of transport and accommodation from The Australian National University and the then Australian Institute of Aboriginal Studies.

Mission staff at Mossman, Mareeba and Wujal Wujal, in particular Mr Ed Stephan and Mrs Lydia Stephan, were most hospitable and contributed a great deal to making my visits successful.

For all this help I am most grateful.

Foreword

Kuku Yalanji, spoken in the area between Mossman and Cooktown in North Queensland, is still a living language. Only about two score of the original 250 distinct Australian Aboriginal languages are still learned by children; Kuku Yalanji is one of them, although its use as the main means of communication in the home has diminished during the past twenty years. This publication is intended to provide a record of the grammar of this language and to make Kuku Yalanji publicly accessible.

This study, based on research undertaken from 1 979 to 1 981, was originally submitted for the degree of Doctor of Philosophy in The Australian National University in December 1 982. (Necessary revision according to examiners' recommendations has been undertaken.) Personal reasons prevented earlier publication; but in the meantime the continued dwindling of Aboriginal language resources on the one hand, and some partial success in a painstaking revival of Kuku Yalanji's southern neighbour Djabugay on the other, led to an increased sense of urgency.

Availability of sound language material has played a role in revival efforts of Djabugay. Reduced to only a couple of fluent speakers by the late seventies, Djabugay has been regaining significance for tribal identity and pride since 1 989 through a language course developed by teacher Michael Quinn in cooperation with Djabugay speaker Robert Banning and other members of the Djabugay community. The language course is primarily based on the sketch grammar written as my Honours thesis in 1 978 and the accompanying taperecorded and written material. Publication of the grammar of Djabugay in 1991 in volume 4 of the Handbook of Australian Languages provided a boost to the language program.

Kuku Yalanji was estimated to have about 300 speakers in 1 990 and it was, and still is, supported by Language Other Than English (LOTE) courses in the Bloomfield River State School and Mossman High School. However, even in the 1 980s it was felt in the community that it was more important for children to speak English at home than Kuku. This makes sense from a pragmatic point of view, but concentration on the dominant language at the expense of the vernacular often creates new problems, namely eventual loss of cultural identity. The effects of the shift towards English as main medium of communication are already being felt. At present not many of the about ninety Kuku Yalanji primary school children are fluent in their language when they enter the LOTE program in grade three.

This grammar is intended to give an accurate and comprehensive description of the components of this language and its grammatical rules to readers with some knowledge of linguistics. While there is some discussion of theoretical issues, the grammar is not cast in a particular theoretical framework; nor does it extend into sociolinguistic analysis. It will serve as a database for the theoretical linguist, or as the necessary background for the scholar

Significant characteristics of this language that are of particular interest to the linguist include:

• vowel-harmony rules,
• two sets of case endings governed by animacy,
• passive and anti-passive constructions,
• treatment of topic and anaphoric devices in narrative discourse.

The language Kuku Yalanji comprised several dialects; only two of these are still in use, Yalanji and NyungkuL This study is concerned with both dialects, recording any differences that have been observed.

1.1 The language and its speakers

1. 1. 1 Kuku Yalanji and surrounding languages

The Kuku Yalanji language is spoken in the south-eastern coastal region of the Cape York Peninsula of North Queensland, Australia (see Map 1 ). The language name can be analysed as 'the language with "this'" (kuku 'word, language', yala-nji 'this-COMIT' ; see §3.2.3.3).I

To the north and south along the Pacific coast Kuku Yalanji is bordered by genetically related languages Guugu Yimidhirr and Dyabugay respectively. On the basis of cognates and particularly of grammatical features one can say that Kuku Yalanji is more closely related to Guugu Yimidhirr than to Dyabugay. Reference to these languages will be made throughout this study, as well as to Yidiny, south of Dyabugay. Genetically, Yidiny and Dyabugay are probably equally close relations of Kuku Yalanji, but because of the greater geographical distance there would have been less areal influence and diffusion between Yidiny and Kuku Yalanji.

Languages in the north-west to south-west of Kuku Yalanji are of a quite different linguistic type. These are: Gugu Warra, Gugu Mini and Mbabaram, an initial-dropping language (see Map 1).

1.1.2 Kuku Yalanji and its dialects

Patz, E. A grammar of the Kuku Yalanji language of north Queensland.

PL-527, xx ^{+ 250} pages. Pacific Linguistics, The Australian National University, 2002. DOI:10.15144/PL-527.1 ©2002 Pacific Linguisticsand/or theauthor(s). Onlineedition licensed ²⁰¹⁵ CC BY-SA 4.0, with permission of PL. ^Asealang.net/CRCL initiative.

The concept of 'own language' as distinct from 'other languages' is an important aspect of tribal identity in Australia, as indeed it is for any nation in the world. Thus Aborigines distinguish their own language from that of another tribal group, even if the differences are only slight, sometimes represented by just a few different lexical items. Accordingly, a tribal name is often derived from the language name in Australia, whereas the reverse is seldom encountered (see Dixon 1 980:3 3-43).

It is not uncommon in Australia that a language name is based on a certain linguistic feature of the language (Dixon 1 980:41-42). Guugu Yirnidhirr, north of Kuku Yalanji, also means 'the language with "this'" (yimi 'this').

Where the self-termed 'languages' of several tribes are mutually understandable, the linguist may recognise these as dialects of one superordinate language. But "there was not usually any name for a group of tribes whose speech was mutually intelligible, nor for the language2 [the superordinate language] that we can - on linguistic criteria - assign to them. It is necessary to choose some label for each language2, and this usually has to be done fairly arbitrarily by the linguist" (Dixon 1 980:43).

Map 1: Kuku Yalanji and its neighbours

Kuku Yalanji has been adopted as the cover term for a language that had about a dozen dialects of which all but two are now extinct or known only to a handful of elderly speakers. The Kuku Yalanji dialect, spoken east of the Great Dividing Range (see Map 2, see § 1 . 1 .3), is one of the two remaining dialects with a viable speech community and its name was originally chosen by Oates and Oates (1 964a) to refer to the eastern dialects of this language group. I extend the reference of this language name to include the western dialects, one of which is also termed Kuku Yalanji by its few remaining speakers. Positive identification and geographical placement of the dialects of Kuku Yalanji is difficult for the following three reasons. Firstly, the name by which speakers refer to their own dialect is not always the same as the term used for it by other communities. Secondly, members of a tribal subgroup may use their territory term as identification (i.e. 'inhabitants of'), which does not necessarily imply a dialect difference. Thirdly, because of the continuous shifting about and intermingling of members of different tribal groups, enforced by European administration over several generations, it may be difficult to ascertain where a particular dialect was originally spoken. Previous research has identified thirteen different dialects for Kuku Yalanji which partly overlap in territory. A full list of these dialects is given in Table 1.1. Data has been collected for those dialects marked (*) in the second and third columns. Map 2 shows the approximate location of these dialects. Several of the dialect names comprise the term kuku (koko) 'word, language' plus presumably a characteristic of the language indicated by the comitative suffix -ji (see §3.2.3.3). However, only two of these are analysable: Kuku Yalanji (see §1.1.1 ) and Kuku Buyunji 'language with bad' or 'bad language'. Other terms cannot be analysed, but it appears that Jungkurara (Tindale column, corresponding geographically to Kuku Tyunkay in Dixon column) may have a territory name as its basis. The ending -ara could well represent the derivation -warm 'inhabitants of' (see §3.2.3.7, and § 1 .2.1.1 'Tribal groups and names'). Possibly, this name does not actually refer to a different dialect (the data for Kuku Tyunkay is in fact identical to that for Muluridyi), but evidence is too scanty to allow a firm decision on this point.

The correspondence in territory for Kuku Buyunji and Yalanji is an example of different terms employed for the same dialect by different groups. Western self-termed Kuku Yalanji speakers (see Brady, Anderson, Rigsby column) refer to the eastern dialect as Kuku Buyunji,2 whereas eastern speakers definitely call their language Kuku Yalanji and refer to the western dialect around Maytown/Palmer River as Kuku Yalaja (see Anderson 1 979:36).

² Haviland (1 979a:29) states that Guugu Yimidhirr speakers use the term Gugu Buyun 'bad language' for an "intermediate dialect with lexical and syntactic affinities to both Guugu Yimidhirr to the North and Gugu Yalanji to the South", that was spoken along the Annan River and was apparently "regarded with disdain by their neighbours".

• Kuku Nyungkul 7 Kuku Jangkun
• 2 eastern Kuku Yalanji (Kuku Buyunji) 8 Wagaman
• (3) (Kuku Tyunkay, Junkurara) 9 Wakura
• 5 Kuku Jakanji 11 Wulbura

• (4) (Kokokulunggur) 10 western Kuku Yalanji (Kuku Yalaja, Kokojelandji)

• 6 Muluriji 12 Kuku Bidiji

Map 2: The dialects of Kuku Yalanji

Tindale (1974)	Brady, Anderson and Rigsby (1980)	Dixon (fieldnotes 1960s)
	Kuku Nyungkul*
Kokobujundji	Kuku Buyunji*	Yalandyi*3
Jungkurara		Kuku Tyunkay*4
Kokokulungkur
	Kuku Jakanji
Muluridji	Muluriji*	Muluridyi*5
Wulpura	Wulbura
Kokobididji	Kuku Bidiji*
Kokowalandja	Kuku Yalanji*
Kokojelandji
Wakara	Wakura*	Wagura*6
Djankun	Kuku Jangkun*	Dyangun (Dyangunbari)*
		Wagaman*7

Table 1.1 : Attested dialects of Kuku Yalanji

H. Hershberger and R. Hershberger ( 1 982) call Kuku Jalunji the "Bloomfield dialect", thus placing it in the territory of Kuku Buyunji or eastern Kuku Yalanji. According to my data, this "dialect", which is spoken by the original seashore inhabitants (jalun-ji 'with sea'), differs from the neighbouring dialects only in that it contains words for maritime fauna and flora that are unknown to speakers living a little further inland. No definite information could be obtained from speakers as to whether Jalunji was actually regarded as "different talk"; but whatever the case may be, Jalunji people were confined to the narrow coastal strip and the term cannot apply to the dialect of the Bloomfield River.

A vailable data, which is only scanty for some dialects, shows that the dialects of Kuku Yalanji are very similar, sharing between over 70% and over 90% of common vocabulary and agreeing closely in morphology·8

The eastern Kuku Yalanji dialect and Kuku Nyungkul, from now on referred to as Yalanji and Nyungkul, are the subject of investigation of this study. (I do not use the name Kuku Buyunji since that is not what the speakers call themselves.) Yalanji and Nyungkul have over

3 Based on Hershberger early 1 960s.

5 Based on Hale notes and recording around 1 960

6 Based on Richards (1924-26).

7 Recorded by Dixon and Laycock ( 1 964).

Tindale also includes a dialect called Wakaman in his map for an area extending a great deal further south than the approximate boundary given by Dixon's informants. According to Sutton ( 1 976a: 1 1 6), Tindale's data corresponds to that of the initial dropping language Agwamin and is clearly not a dialect of Kuku Yalanji. Apparently, the name Wakaman was used by different informants for different languages (see Sutton 1 976a and Dixon field notes).

8 The small corpus of western Yalanji data obtained from Norman Mitchell shows some differences from eastern dialects in verbal morphology. Since the data is very scanty no further comments are included here.

⁴ Based on Hale field notes around 1 960.

90% of common vocabulary and show only minor morphological differences. Because of the attested close similarity between the dialects of Kuku Yalanji, the linguistic details described here may be taken to apply to the Kuku Yalanji language. Example sentences, unless otherwise indicated, are given in the Yalanji dialect and special mention is made of morphological differences in Nyungkul.

1. 1.3 Territory

Speakers of the Kuku Yalanji language traditionally inhabited an area of over 2000 sq km extending from the Mossman River in the south to the Annan River in the north, bordered by the Pacific Ocean in the east and extending inland to presumably just west of Mount Mulgrave (see Map 2). Three ecological zones may be distinguished in this region: open woodland, dense rainforest and coastal plains.

To the west of the Great Dividing Range there is mostly flat to hilly country with open woodland and grass vegetation. Although there are a number of larger rivers such as the Palmer, Mitchell and Saint George Rivers with their tributaries, these rivers are mostly seasonal and may dry up to only a few waterholes in the dry season. Nevertheless, in precontact times this area would have been able to support a fair number of people with good resources of game, fish and plant food.

The eastern slopes of the Great Dividing Range leading down to the Pacific coast are extremely rugged and covered with dense rainforest containing a multitude of permanent springs, creeks and rivers, which provide a reliable water supply and fish resources. Game and plant food were also abundant in the rainforest region. For the most part the slopes descend steeply right to the ocean front, creating a continuous chain of small bays with narrow sand beaches.

More extensive coastal plains exist on the mouths of the larger rivers, e.g. the Bloomfield River and Daintree River. Here the vegetation consists largely of coastal thicket and mangrove swamps. The coastal waters and estuaries had plentiful and reliable resources of fish and other edible maritime fauna, and coastal vegetation also provided some plant foods. These coastal plains and the adjacent gradually narrowing river valleys were apparently heavily populated in pre-contact times, as may be inferred from Hughes' list ( 1 886) of no less than ten tribal groups in Weary Bay at the mouth of the Bloomfield River.

The Yalanji and Nyungkul groups traditionally inhabited the rainforest terrain (see Map 3) in which the present Wujal Wujal Community is situated.

1.1.4 Present speakers

The devastating effect of the Palmer River gold rush late last century very quickly decimated the Kuku Yalanji tribes west of the Great Dividing Range to such an extent that several of them are now extinct. It is difficult to estimate the number of remaining speakers of any western dialect, since they are widely dispersed over an area from Mareeba to Cooktown, Kuranda, Chillagoe, Mossman and the Bloomfield River. Others have been resettled at Yarrabah or Palm Island. Three elderly speakers that I briefly consulted were very proficient in their language (western Kuku Yalanji) but bemoaned the fact that the younger generations are on the whole ignorant of the language of their forebears. There is certainly nothing remotely like a 'speech community' for any of the western dialects.

Map 3: Nyunkul and Yalanji Territory

Eastern Yalanji and Nyungkul are still spoken by about 300 speakers (Schmidt 1 990). Owing to the ruggedness and inaccessibility of their territory they remained largely unaffected by the violence accompanying early European intrusion and have managed to maintain many aspects of their culture and particularly their language. A number of Yalanji and Nyungkul speakers now live at Mossman, but the majority are concentrated in the Bloomfield River valley. There are slightly more Nyungkul than Yalanji speakers and there is also a very small proportion of Bidiji, Yalaji and Guugu Yimidhirr speakers in this area.

Language consciousness and maintenance on the Bloomfield River was boosted by the introduction, in the late 1 970s, of a vernacular literacy program in Kuku Yalanji. The program was developed by the SIL missionary linguists H. Hershberger and R. Hershberger and taught by them, assisted by Aboriginal teacher aides, at the Bloomfield River State School until their departure in 1 985. Aboriginal assistant teachers have been teaching Kuku Yalanji as a Language Other Than English (LaTE) from preschool to year 7 until recently when the program was reduced to begin in year 3. Not only Aboriginal children - in 2001 about 85% of the student population - but also non-Aboriginal children take part. The teaching material is based on the Hershbergers' program, but the school has been updating and revising the material (see Harms et al. 1 997). The Mossman High School, too, includes Kuku Yalanji in its LaTE program for Grade 8 students. It appears that within the constraints of available funding both schools are contributing a great deal towards the maintenance of Kuku Yalanji and its associated culture. However, with two half-hour lessons per week in the primary school program the impact cannot be very strong.

By 1 981, according to my observations, the main language of communication among Aborigines at Wujal Wujal was KukuYalanji and Kuku Nyungkul. The same seemed to apply, at least among the adult population, when I visited the community briefly in 1 990. However, even 20 years ago several parents expressed their view that it would be advantageous for their children's school education to speak English in the home. They therefore did not want to enforce the speaking of Kuku Yalanji in the family. Apparently this trend has continued; a Bloomfield River State School teacher reported in 1 993 that children understood Kuku well, but many were less competent at speaking it. In 200 1 Norman Tayley, Wujal Wujal Community Council member and assistant teacher, confirmed that the use of Kuku in the home had further decreased, that it was hardly used in the community's kindergarten and that consequently the children did not have a proper grasp of the language.

A form of creole is also in use. Typical creole features that I noticed were: suffix -im on English transitive verb roots, e.g. put-im, chuck-im; suffix -bala on adjectives, e.g. that deadbala 'that dead person',jlash-bala 'distinguished' (a slightly derogatory term); blonga or bla 'to indicate possession', e.g. car blonga/bla Cedric 'Cedric's car'; and lexical items too much and big mob for 'a lot', savvy for 'know'.9 For some older speakers the creole is the only way to communicate with white Australians.

1. 1.5 Studies of KukuYalanji

Cursory investigations of several dialects obviously belonging to KukuYalanji have been preserved from the period of European settlement between the 1 870s and about 1 900. Word

⁹ For a detailed discussion of Cape York Creole, see Crowley and Rigsby ( 1979).

lists varying in length and quality are available for tribes living around the headwaters of the Hodgkirison, Mitchell and Walsh Rivers, presumably speakers of Wakura (Richards 1 924; Mowbray 1 886; Davidson 1 886). Vocabularies from the inhabitants of Weary Bay and the Bloomfield River valley were collected by Hislop ( 1 899), Hughes (1 8 86) and missionary Hoerlein (unpublished, presumably around 1 900). Hislop, one of the first settlers in the Bloomfield River valley and a sensitive observer of the Aborigines (see Roth 1 901-1905) provided an excellent word list which agrees closely with the present data for Yalanji and Nyungkul. Hoerlein's vocabulary with about 550 entries is also of good quality and concurs with present data from the area.

Schmidt, in his Die Gliederung der Australischen Sprachen ( 1 919), used Hislop and Hughes as sources, calling the language Bulponarra, which is the first of the ten names for Weary Bay tribes recorded by Hughes. (Lanyon-Orgill 1 962 gives a list of fifty words of the Bulponara language, allegedly collected prior to 1 80 1 ; however, like much of Lanyon-Orgill's other work, this is probably a hoax.)

Except for some brief investigation of an inland dialect by Tindale in the 1 930s, no further research of the Kuku Yalanji language was undertaken for almost sixty years. Hale, Dixon, and Laycock and Dixon collected some material on several dialects in the early 1 960s (see Table 1. 1 ).

In the late 1 960s and early 1 970s some material of "Gugu-Bujun and Related Languages" and "Kantyu and Koko-Yalandji" was collected and described by N. Holmer and published in 1 988. Holmer's description and analysis of the data contains errors, inconsistencies and unsupported statements. Only a few examples will be given here; for more detail see Dixon ( 1 992). In his chapter 'Gugu Buyun and related languages' Holmer (1988:38) describes in very broad but basically accurate terms the vowel-harmony rules applying in Kuku Yalanji (see §2.5.1). Later, in his chapter 'Kantyu and Koko-Yalandji' he describes the same vowel-harmony rule but extends it to monomorphemic words and explains "the appearance of -i for expected -a" as being "due to the phonetics of Queensland English" (Holmer 198 8 :65). He includes a retroflex stop and retroflex nasal in the consonant system, neither of which in fact exists in Kuku Yalanji (see §2. 1 . 1 ). His classification of locative suffix allomorph -ba as a "postposition" is unsupported ( 1 988 :39, 66) and his identification of stem-forming suffixes "personal-man" and "feminine -gan" is wrong. In short, Holmer's comments do not contribute to the understanding of Kuku Yalanji.

Detailed investigation of Yalanji and Nyungkul began in 1 961 with H. Hershberger and R. Hershberger of the Summer Institute of Linguistics. Between 1 964 and 1 982 they published sketches of the following aspects of Yalanji/Nyungkul: pronouns, nominal and verbal morphology, structure of noun phrases, clauses and sentences, two clitics, stress patterns (all using the tagmemics approach). Several of these publications had been prepared after only six months of fieldwork and are inevitably sketchy and subject to some misinterpretations. In 1 982, with considerably more familiarity with the language, a twoway dictionary appeared with close to 2000 entries, some example sentences and a very brief non-technical grammatical outline. In 1 985 their translation of the New Testament was published.

W. Oates and L. Oates (1964), also of SIL, published notes on phonology, pronouns and cultural background, as well as a two-way vocabulary with about 1000 entries, apparently after only two months of research. The quality of the vocabulary leaves something to be desired, particularly the representation of verbs, which are often listed not as roots but indiscriminately in different inflectional forms. In 1 992, L. Oates published a Kuku Yalanji

dictionary based on the Oates 1 964 vocabulary, the Hershberger dictionary and "two handwritten word lists by Norman Baird", Aboriginal Assistant Teacher at Bloomfield River State School. The dictionary, arranged by semantic fields, is apparently designed for teaching purposes. It includes instructions on morphology and grammar and sample sentences, many of which are taken from Hershberger's translation of the New Testament.

A detailed discussion of suffix -ji was undertaken by 1. A. Ware in 1981 for a MA thesis at North Texas State University. Based on Oates and early Hershberger data Ware argues that -ji, both as nominal and as verbal suffix, is basically a stative marker. While some functions of verbal suffix -ji support this notion when seen in the context of transitivity hierarchy, the range of functions is much more complex (see §3.8.5.4 (ii) and §4.2.4). But more importantly, verbal -ji and nominal -ji, which is just one allomorph of the comitative suffix (see §3.2.3.3), are not the same suffix. Ware's analysis apparently suffers from the limitations of the data available to her.

The work by Hershberger and Oates prior to 1 982 was used as a starting point for this study and has been amended where indicated.

My own investigation of the Yalanji and Nyungkul dialects is based on about 10 hours of recorded texts (85 texts from one to fifteen minutes in length) and about 19 hours of recorded elicited material; further elicitation was taken as notes but not recorded on tape. (Tapes and written material are deposited with the Australian Institute of Aboriginal and Torres Strait Islander Studies in Canberra.) Further insight into the language was gained from listening to and taking part in casual conversations.

Almost all of my data was gathered from speakers of the Wujal Wujal community where I spent about ten months of fieldwork on four visits during 1 979-8 1, becoming an active participant in community life. A small portion of data was collected from Yalanji and Nyungkul speakers living in Mossman and from a speaker of an inland dialect, presumably Kuku Yalaja, living at Mareeba.

In the Wujal Wujal community I worked mostly with a permanent group of five Yalanji and Nyungkul speakers in the age group between 45 and 65. These speakers proved to be most reliable both in their linguistic competence and ability to express language intuitions, and in their wealth of traditional knowledge.

1.2 The culture of the Kuku Yalanji

It can be assumed that there was reasonable cultural homogeneity in the area specified for the KukuYalanji language (see Anderson 1 980:78). However, post-contact history took a dramatically different course in the western and eastern parts of this territory and first-hand information on cultural background is almost exclusively confined to the eastern speakers of KukuYalanji who represent a still largely intact community.

1.2.1 Cultural background

This subsection can be no more than a cursory account of some aspects of Kuku Yalanji culture. Unless otherwise indicated, the details presented here are based on personal information from members of the Yalanji and Nyungkul community. (For more detailed discussions of social and particularly economic patterns see Anderson 1 979 and 1980.)

1.2. 1. 1 Tribal groups and names

The language name may be used in unchanged form to refer to a native of a Kuku Yalanji-speaking area. As is characteristic of Australian Aborigines, Kuku Yalanji society is closely associated with the land to which they belong. Thus a distinction is made between six groups or 'mobs' which "were and are identified with particular territorial tracts focused more or less on river drainage areas" (Anderson 1 980:78). Members of different 'mobs' used to congregate for ceremonies or when a particular food was abundant in a certain area. They still do so today for festivities that either are arranged spontaneously or are linked to the European calendar, such as Christmas.

Within each 'mob' there are several descent groups associated with "patrilineally inherited relatively boundary specific 'countries' or bubu" (Anderson 1 980:78). These groups are known by the name of their 'country', e.g. banabila-warra 'inhabitants of Banabila (near the mouth of the Bloomfield River)', or wujalwujal-warra 'inhabitants of Wujal Wujal (the mission site on the Bloomfield River), (see Hughes' 1 886 list of tribal names for Weary Bay). The descent group is obliged to care for and maintain their 'country', which usually contains one or more sites of spiritual or mythological significance. These sites are called yirmbal 'story place'.

An overall distinction which encompasses 'mobs' and descent groups is made according to ecological features of the territory, e.g. majaji 'with rainforest' versus ngalkalji 'with open woodland' for rainforest and woodland dwellers, andjalunji 'with sea' versus bana yiralji 'with fresh water' for seashore and river dwellers.

1.2. 1.2 Personal names

All members of the Wujal Wujal community now have English first names and surnames. Most people over about 60 years of age also have Aboriginal names, but they are reticent about disclosing them and people are rarely called openly by their native name. Only some native language nicknames are in general use, e.g. jina-baji 'sore foot' for a person who had his foot injured by a stonefish when a young boy.

1.2. 1.3 Social structure and marriage

Each Yalanji and Nyungkul speaker belongs to one of two moieties: walarr or dabu, both also referring to bee species. The main totems of these moieties are birrmba 'white cockatoo' and kumurrbina 'black cockatoo' respectively (see W. Oates & L. Oates 1 964a).

A correct marriage partner belongs to the opposite moiety, and children belong to the same moiety as their father. Owing to an overall decrease in number of the KukuYalanji people and earlier interference by missionaries, as well as a somewhat more liberal point of view among the majority of the younger generation (but by no means all of them), it is not always possible nowadays for a 'right' marriage to be arranged. However, if a marriage turns out not to be a happy one, this is generally blamed on the fact that the partners are not 'right' according to tradition.

Formerly, a man could have as many wives as he could support, bearing in mind that he also had to provide to some extent for his in-laws.

12 Chapter 1

Each family member had an intrinsic right to a particular portion of the available food which ensured that provision was made for those who could not fend for themselves, such as the elderly or sick. This convention of obligatory sharing has been extended to money supplies and is still adhered to in the present community (but unfortunately it is now often subject to abuse, according to my informants.)

The relationship between a man and his mother-in-law was and is subject to total avoidance, that is they may not speak to each other or even be in the same room. Less strict taboos apply to other in-law relationships, where the persons concerned may converse but not, for instance, touch each other. These taboos have created some problems in the running of the community, since for instance the local buLiman 'policemen' will refuse to take, say, a drunken father-in-law into custody, and a mother- or son-in-law will miss out on her or his old age pensioner meal, because they must not be together in the room where the meals are served.

There is evidence that an avoidance language was formerly used among certain in-law relations, although it is not clear to which relationships this applied. The last person to know this avoidance style died many years ago and no records of this speech style exist. ¹⁰

1.2. 1.4 Food

KukuYalanji territory on the whole had ample food resources. Particularly the rainforest region and coastal strip provided abundant and reliable food supplies, such as fruit and nuts (e.g. bush cherry, blue quandong, wild apple, figs, zamia nut, Moreton Bay chestnut), roots (e.g. yams and wild ginger), wild honey, smaller mammals (e.g. possums and bandicoot), birds and eggs (e.g. scrub hen, cassowary, Torres Strait pigeon), goannas, pythons (poisonous snakes were not eaten), fresh- and saltwater varieties of turtles, fish, and shellfish, crustaceans and insects (e.g. green ant, witchetty grub) (see Anderson 1 979; Roth 1 90 1- 1 905 and 1 907-1910).

Fish, shellfish, turtles and scrub hen eggs as well as some plant foods are still in relatively good supply and are often obtained to supplement the European diet (see Anderson 1 980). A new 'natural' food resource is mangoes and wild pigs, whereas cassowaries, native mammals and pythons have all but disappeared since the introduction of the poisonous cane toad in the area.

Preparation of some plant foods such as zamia nut and hairy yam required elaborate preparation, including roasting, pounding into flour and subsequent leaching in water for several hours (see Roth 1 90 1- 1 905 and 1 907-1 910). In the case of zamia nut, leaching is required to extract the poisonous substance, whereas hairy yam needs to be soaked in water to remove the unpleasant taste. Such labour-intensive foods are rarely prepared today; they have been replaced by white bread and damper.

Meat and fish used to be roasted in a ground oven between hot stones and various types of leaves (see Roth 1 901-1905 and 1 907- 1 910, see also Text 51 in the appendix) and this is still a popular and tasty way of cooking, particularly for freshwater eels, during 'walkabout' expeditions. Boiling in a bark trough or a large shell was a traditional mode of cooking food in the Bloomfield River region but apparently unknown anywhere else in North Queensland.

¹⁰In the closely related language Guugu Yimidhirr an avoidance speech style is still used between brothersin-law. A description of this can be found in Haviland (1979a, 1979b). However, it would be mere conjecture to assume that the Kuku Yalanji avoidance style was in any way similar.

Cannibalism was not unknown among the Kuku Y alanji, but was only practised when people were driven by hunger. The victim was not killed for that particular purpose, but was eaten after death by natural causes or in a fight. As Roth (1901 :30) reports:

... there was no special distinguishing term for cannibalism, which had no special ceremony attached to it, was not confined to any particular members of the tribe, and was only resorted to when compelled by hunger. There was, however, one instance recorded here [in the Bloomfield River District] where the natives ate the killed on the battle field, both sides joining in the repast, and subsequently resuming the fight.

On the Palmer River, where the huge influx of European and Chinese miners during the gold rush quickly destroyed woodland, rivers and waterholes and with them the resources of game and fish, the Aborigines were soon forced to find alternative food supplies in the form of the miners' animals and the miners themselves. Stories of cannibalism on the Palmer River abound and my informant Charlie Tayley was able to tell me (from hearsay) that Chinese were regarded as better eating than Europeans and that a slain enemy was further utilised by fashioning his skull into a billycan (see Anderson & Mitchell 1 981).

1.2. 1.5 Artefacts

Different types of dillybags and traditional hunting weapons such as fish and turtle spears are still made and used today, but metal has replaced the traditional materials for spear prongs and hooks.

1.2. 1.6 Medicine

Minor ailments were and largely still are treated with bush medicine, such as yangka 'green ant Guice)' for fever, colds and stomach upsets, or poultices from stewed bark of kabal 'Leichardt tree' or leaves of jujubala 'ironwood tree' applied to a limb for rheumatic pain. In severe or chronic cases of illness the rrunyuji or murri doctor was called upon. He would extract 'by magic' a 'thing', usually a stingray barb or in later times a marble, from the patient's body. This 'thing' was believed to be the cause of the illness and the patient was expected to get better after its removal. One of my informants claimed that some years ago she had been cured in this way from an abdominal complaint after unsuccessful treatment by European doctors. A murri doctor with such healing powers was said to live in Cooktown in the early 1980s and was still consulted as a last resort if western medicine proved ineffective.

1.2. 1.7 Taboos

Taboos applying to in-law relationships have been mentioned in § 1 .2. 1 .3 above. Food, either from certain places or obtained by particular family members, may also be subject to taboo. Any game or fish from the north-eastern part of Weary Bay, for instance, is believed to bring about severe illness and must not be eaten. My informants repeatedly warned me against fishing in this area, illustrating their argument with examples of cancer and paralysis that had befallen both Aborigines and white settlers who had not heeded this taboo.

14 Chapter 1

Food collected by a family member is taboo to certain siblings. For instance, a woman is not supposed to eat meat or fish obtained by her younger brother, and a man should not eat certain plant foods and eggs gathered by his younger sister. This rule is still largely observed, whereas another taboo that forbade certain types of game for children and adolescents does not apply any more. (As one of my informants put it: "The old people just wanted to keep the best food for themselves".)

Mention of the name of a deceased person was proscribed totally in former times for as long as the body was decomposing. (The corpse was wrapped in bark and placed in a tree under which a fire was maintained until "only the bones were left". These were then carried around for a while by family members before being buried.) During the decomposition period the name was replaced by the substitute term ngalba in reference to the deceased as well as living persons with the same name. !! Nowadays, the substitute term plus the name, e.g. ngalba 'Ivy', is generally used, whereas family members refer to the deceased only by a kinship term. A recently deceased person is talked about as little as possible and tape recordings, for instance, of such a person must on no account be listened to.

1.2. 1.8 Other customs and beliefs

Initiation practices such as piercing of the nasal septum, front tooth extraction and 'marking' with cicatrices have gradually been discontinued over the last sixty years or so, incidentally much to the relief of those people who just managed to avoid such painful operations in their youth.

Persistent offenders, e.g. thieves or breakers of sexual or other taboos, used to be speared, sometimes even to death, according to the severity of their crime.

A common 'magic' practice was the kanyil 'curse song'. The last (reported) kanyil was performed in the late 1 930s. If a person was wronged by someone else, for instance in cases of adultery, he/she would 'sing' the offender(s), which was believed to ultimately bring about the culprit's death (see W. Oates & L. Oates 1 964a).

A person's death was often thought to be caused by sorcery. It was believed that the dead person's belongings would attract the 'spirit' of the sorcerer. This spirit could only be seen by the rrunyunji who would hold watch until the murderer's spirit appeared. Once the murderer was identified, the dead person's family had the duty to avenge their relation by spearing. (Oates and Oates 1 964a report a different way of detection by pointing a stick.) This type of magic and punitive measure has not been practised for many years.

A source of great fear even today are thunderstorms, which usually begin in early November and increase in frequency and intensity until the beginning of the wet season towards the end of December. Some actions are thought to bring about a thunderstorm, e.g. trespassing on a sacred site. Certain practices were and are believed to drive away the storm, such as burning ironwood leaves or turtle skin.

Some 'messenger' birds are accorded great importance, as for instance wakuka 'kookaburra', who is thought to herald the death of a family member.

¹¹Ngalba is not etymologically transparent in Kuku Yalanji, but it is presumably cognate to Guugu Yimidhirr ngalba- 'hide, be hidden'. The use of a particular word that is reserved for the sole purpose of substituting a tabooed name is attested only for central Australia, western Queensland and Cape York Peninsula. For a discussion of this strategy in central Australia, see Nash and Simpson (1981).

One of the spirits still believed in is wuyngkul, the spirit of a sick person close to death, who moves about intent on mischief such as causing accidents or sudden pain. It is visible to specially gifted people and will disappear again on the death or recovery of the sick person. During my stays on the Bloomfield River several older people attributed minor accidents or pains to a wuyngkul that was thought to be abroad.

Creation myths of Yalanji and Nyungkul speakers centre on mythical beings which have become associated with certain geographical features. Some of these are kija 'moon', who created women at kijandaku 'Roaring Meg Falls'; kurriyala 'carpet snake, ' who died and is buried at mangkal 'Cedar Bay'; ngurrku 'mopoke'; and bajabaja 'blue tongue lizard', whose eggs hatched in the Bloomfield River bed, dry at the time, and later turned into two prominent rocks in the river. The origin of all rivers and creeks in the area is thought to be the amniotic fluid of one of two sisters who gave birth on ngalbabulal 'Mount Pieter Botte', about halfway between Bloomfield and Daintree.

A spirit which is associated with one particular Yalanji family clan is wadalwadal, a water spirit with female body and fishtail, living in a spring in the Thompson Creek valley. It will not suffer trespassers on its site, but tolerates members of the family who rightly belong there and who may use the spring water that is believed to have healing powers.

1.2.2 Recent history

The earliest mention of Aborigines in Weary Bay was made by the explorers Cook ( 1 770) and King ( 1 819, 1 821) who noticed signs of habitation such as campfires and canoes, but saw no people. The first inland explorer William Hann reached the Bloomfield region in 1 872 and reported several encounters with Aborigines, which were generally friendly.

From 1 873, with the discovery of gold on the Palmer River and later tin in the Annan River region, European migration into KukuYalanji territory increased suddenly and dramatically, but its impact was markedly different on the eastern and western tribal groups.

West of the Great Dividing Range where the Palmer River and Hodgkinson gold fields attracted thousands of European and Chinese miners, the Kuku Yalanji fought a fierce but losing battle for survival. Alluvial gold mining in the rivers deprived them of their supplies of water and fish, and game resources were quickly depleted by the environmental impact of the sudden huge alien population with their horses and cattle. Hostile encounters predictably resulted in the death of large numbers of Aborigines and the few successful attacks by the natives had no significant effect other than leading to punitive measures by the Europeans. The open country afforded little protection for the Aboriginal population, who was soon reduced to a few remnants.

East of the Great Dividing Range the Aborigines stood a much better chance of survival for two reasons: firstly, the main attraction of gold was lacking in this region and the influx of Europeans was only slight compared to the Palmer River area. Secondly, the rugged mountainous country covered with virtually impenetrable rainforest prohibited any significant spread of European population while at the same time providing the Aborigines with reliable water supplies, largely undisturbed food resources and, not least, possibilities for concealment. As Anderson (1 979:34) states:

Because of the relative isolation of the Bloomfield River district and due to its extremely rugged terrain and dense vegetation, encounters between Aboriginal people and white settlers were minimal and of a relatively peaceful nature,

16 Chapter 1

and ( 1 979:36):

Due to the nature of the Bloomfield terrain and the types of European activity there, Aborigines could more or less control the amount of interaction they had with whites.

In the late 1 880s there were about 1 200 Europeans in the area between the Annan and Daintree River, mostly tin miners, but also Red Cedar loggers and settlers involved in agricultural enterprises such as sugar, tobacco and coffee plantations. None of these encroached significantly on Aboriginal territory.

Relations between Aborigines and Europeans in this area gradually developed into some form of mutual assistance. Europeans recognised a source of cheap labour in the Aborigines and the latter in turn were quite partial to the commodities that they received as payment, e.g. tobacco, flour, sugar, and tea, and even came to rely to some extent on European food supply. Reliance on foreign food developed mainly because Aboriginal males working for Europeans often spent time away from their tribal groups which deprived their families of their main providers of game and fish. To compensate for this, women often agreed to sexual relations with nearby white miners or loggers who would provide them with food for their families. This was usually done with the consent of the Aboriginal husbands (see Anderson 1 979:35).

Later on, beche-de-mer and pearl shell fishing off the east coast also used Aboriginal labour, but recruiting practices for these enterprises were often of a dubious nature. Boys and young men were either kidnapped by white fishers or sold off by older men of their tribes (see Anderson 1 979:35; J. Haviland and L. Haviland 1 980: 1 34; Loos 1 980).

In 1 8 86 the Queensland Government gazetted a 'Reserve for Aborigines' of about 640 acres as well as a hunting reserve of about 30,000 acres, against the opposition of local settlers who felt that the land could be more gainfully utilised. (An assumption which incidentally was proven wrong by history, since all larger agricultural enterprises in the area failed owing to poor soil and the difficulty of marketing produce because of transport difficulties from this isolated area.)

In 1 887 the Lutheran church was given control of the reserve in an attempt to settle and 'civilise' the Aborigines through regular employment and schooling, but this attempt failed and the mission was abandoned in 1 902. The mission site had proved unsuitable for agriculture and Aborigines never remained on the mission for long. They preferred to control their own lives, which was possible partly through traditional subsistence patterns and partly through symbiotic relationship with European settlers, who did not require the commitment that was expected by the missionaries. There is even evidence that Aborigines, being aware of their own value as labourers, played off the Europeans against each other (Anderson 1 979:36).

For the following fifty years Yalanji and Nyungkul people suffered relatively little interference (compared to other regions in Queensland) from their 'protector', the Queensland Government. Throughout this time Aborigines always slightly outnumbered the European population and continued their assimilation of tribal and European subsistence patterns, congregating in several major camps usually in the vicinity of European settlements. After World War II they moved, or were moved, into the greater Bloomfield valley where a mission station was reopened in 1 956 under the administration of the Queensland Lutheran Mission Board. By 1 988 the population of Wujal Wujal numbered about 3 30.

Since the 1 950s three factors have had an increasingly detrimental effect on the wellbeing of the community: traditionally antagonistic tribal groups now living together, a great reduction in accessible tribal land and lack of self management. 12 In March 1 987 a young man killed himself while detained in the Wujal Wujal watchhouse. The resulting inquiry as part of the Royal Commission into Aboriginal Deaths in Custody recounted a disturbing story of an Aboriginal community under threat. Access to traditional lands had been greatly curtailed because of "tourist development, real estate speculation, mining and commercial fishing industries and conservationist interests" (Wyvill 1 990:9). Tensions within the community could no longer be alleviated by spreading out on socially productive hunting, fishing and food gathering trips but were acerbated by poor and cramped living conditions, unemployment, alcohol abuse and a general feeling of boredom and frustration (see Wyvill 1 990).

It is therefore heartening to find that at the turn of the millennium Wujal Wujal, now governed by its Aboriginal Community Council, is a growing and apparently optimistic community. Community Development Employment Projects (Commonwealth Department of Education, Training and Youth Affairs) provide employment opportunities; the Women's Centre produces arts and crafts for the tourism trade; the Wujal Wujal Kindergarten has about ten children in the care of Aboriginal teachers. Also, the fact that an illegal alcohol outlet close to the 'dry' community disbanded some years ago can only be beneficial.

1.2.3 Summary

Concluding the observations on cultural background and recent history one can say that, while the western KukuYalanji succumbed early to the sudden and massive foreign immigration into their territory, contact in the eastern parts progressed more gradually and relatively non-violently. Initially, this allowed the native inhabitants, Yalanji and NyungkuI, the opportunity to assess and adopt those aspects of European civilisation that appeared desirable and to adjust slowly to a western lifestyle, while at the same time maintaining for a long time much of their cultural integrity based on traditional values, beliefs and subsistence patterns. Developments in more recent years have shown that traditional ways were declining, primarily because of lack of land. It also appeared that white administration of the mission settlement had left the Kuku Yalanji and Kuku Nungkul ill-equipped to cope with the now rapidly changing living conditions. However, most recent developments seem more positive and hopeful with respect to the survival of a viable Aboriginal community on traditional land, albeit very much reduced in area, within the framework of mainstream Australian society and economy.

2.1 phoneme inventory

Kuku Yalanji has an inventory of sixteen segmental phonemes, consisting of thirteen consonants and three vowels, which is one of the smallest phoneme inventories found in Australian languages.

2. 1. 1 Consonants

Kuku Yalanji consonants are shown in Table 2. 1 represented by IPA symbols as well as the letters used by H. Hershberger and R. Hershberger in their practical orthography for Kuku Yalanji, which will be used throughout this text. The practical orthography devised by the Hershbergers is used in the vernacular program in the Bloomfield River State SchooL There are also a number of SIL publications produced for the community that use this orthography, e.g. booklets on child-care, nutrition etc. In my description of Kuku Yalanji grammar I feel obliged to adopt the symbols that Kuku Yalanji speakers are already familiar with. However, I have made two changes in the spelling of some words. The Hershbergers do not distinguish between final n and ny, representing both sounds as n, because they wanted to avoid confusion with the English spelling and pronunciation of final ny as in many. But since final n and ny are clearly distinct (see minimal pairs further on in this sub-section) I do use both symbols word-finally. I also use final y after vowel i (see §2.S.2).

The practical orthography includes one apparently inconsistent aspect. Stop sounds are usually voiced and bilabial and alveolar stops are appropriately represented by letters b and d respectively. The velar stop, on the other hand, is represented by letter k. H. Hershberger and R. Hershberger chose this letter to make an easy distinction between the velar nasal ng and a sequence of alveolar nasal plus velar stop nk.

Stops and nasals in Kuku Yalanji include two peripheral (or non-coronal) series, bilabial and dorso-velar, one apical series and one laminal series. Bilabial stops and nasals are made simply with the lips. The apical series involves the tip of the tongue touching the alveolar ridge as for English [d] and [n]. Stop /d/ may be realised intervocalically as a tap and some speakers have occasionally even been recorded as using a trill [r] in this position. Two words where a trill instead of stop /d/ has been used with some consistency by certain speakers are:

karray (kaday)	'come'
barriy (badiy)	'cry'

	bilabial	apico- alveolar	apico-post alveolar	lamino- palatal	dorso- velar
Stop	b [b/p]	d [d/t]		j [/c]	k [g/k]
Nasal	m [m]	n [n]		ny [ɲ]	ng [ŋ]
Lateral		l [l]
Rhotic		rr [r]	r []1 (retroflex)
Semivowel		w [w]			y [j]

Table 2.1: Kuku Yalanji consonants

In a vocabulary list compiled by missionary Hoerlein around the tum of the century the word 'come' is clearly recorded with a stop: katay. (,Cry' is not included in this list.) On the other hand, note that cognates for 'come' and 'cry' in Dyabugay, the southern neighbour of Kuku Yalanji, both contain a trill: karray, barriy (Patz 1 991), whereas Yidiny, Dyabugay's close genetic relative, has badi- and gada- (Dixon 1 977a), and Guugu Yimidhirr, to the North of Kuku Yalanji, has baadhi- with a lamino-dental stop, and gadaa- (Haviland 1 979a). The trill variant of Idl in Kuku Yalanji has been recorded most consistently among speakers who grew up or still live in Mossman, the traditional boundary area, where speakers of both Kuku Yalanji and Dyabugay had been moved together on a mission station. Thus it may be possible that Dyabugay pronunciation has had some influence on these cognates in Kuku Yalanji. On the other hand, and I regard this as the more likely possibility, Id/ and IITI could be merging in intervocalic position in Kuku Yalanji. Tap [r] (see below) is an allophone of both stop Idl and trill IITI in intervocalic position in present-day Kuku Yalanji. The next stage could be the reinterpretation of intervocalic [r] as trill, a phonetic change which may already have been completed in Dyabugay.

Laminal stop and nasal, Iji and Iny/, are produced by placing the blade of the tongue against the hard palate. The laminal stop is usually realised as an affricate in all environments.

Dorso-velar consonants are realised by the back of the tongue being raised against the soft palate as for English g/k, and ng as in sing.

The following minimally distinct words illustrate the phonemic status of stops and nasals in different places of articulation:

Stops

bilabial	bubu	'land, place'
apico-alveolar	dudu	'blunt'
lamino-palatal	juju	'bum'
dorso-velar	kuku	'word, language'

See Ladefoged (1 975) Consonant Chart.

Nasals

bilabial	bama	'Aborigine'
apico-alveolar	bana	'water'
	jalban	'tree top'
	nandal	'close, cover'
lamino-palatal	jalbany	'taboo food'
	nyandal	'chop'
dorso-velar	ngandal	'mouth'

Stops are always unaspirated and there is no phonemic distinction between voiced and voiceless. There is a general tendency for stops to be voiced intervocalically and in consonant clusters following /l/-, retroflex /r/- and nasals, while in other positions they may be voiceless.

Like other languages east of the Gulf of Carpentaria (Dixon 1980:143) Kuku Yalanji has only one lateral, apico-alveolar /l/, which is pronounced with the tip of the tongue touching the alveolar ridge. After vowels /a/ and /u/ the lateral is slightly velarised, producing a darker lateral sound similar to [ɫ] in English table, for example:

	kalal	[1]	'vine'
but	milbi	[1]	'show'

The two rhotics in Kuku Yalanji are distinguished by place and manner of articulation. For trill Irrl the tip of the tongue vibrates against the alveolar ridge while the retroflex continuant Irl is produced by slightly turning back the tip of the tongue and approaching the roof of the mouth with the underside of the tongue tip. Its pronunciation is similar to the continuant in American 'bird' but without lip rounding. Trill Irrl is usually reduced to a tap [f] in intervocalic position in normal speech (but not in citation forms), whereas it is clearly trilled in other positions, for example:

	marra	[r] or [r]	'zamia nut'
	jirray	[r] or [r]	'much, plenty'
but	ngurrku	[r]	'owl'
	ngulkurr	[r]	'good'.

The following minimal pairs illustrate the phonemic status of both rhotics and the lateral:

malal	'spider'
malar	'gooseflesh'
maral	'girl'
mararr	'sail'
marral	'dry'
karrarr	'seaweed'
karrar	'species of bird'
kararr	'grass, blanket to lie on'

The Kuku Yalanji phoneme inventory also contains the two semivowels labial-velar Iwl and palatal Iy/. While semivowel Iwl and vowel lui on the one hand and semivowel Iyl and vowel Iii on the other hand are regarded as articulatory identical pairs, semivowel and vowel

are distinguished by duration. A vowel has continuing articulation, whereas a semivowel is characterised by momentary articulation which is produced by a quick movement towards and/or away from the corresponding vowel-type stricture (Catford 1 97 7:131, 165). Semivowel /w/ before or after vowel /u/ is acoustically distinct from the vowel, due to a higher degree of lip rounding and presumably also a narrowing of the cross-section between the articulators in the velar region. Semivowel /y/ before or after vowel /i/ is distinct from the vowel only by a narrowing of the articulatory cross-section by raising the anterodorsal part of the tongue. (Strictly speaking, semivowel /y/ is not a lamino-palatal sound, since a section of the tongue just behind the blade area is involved in articulation. However, to distinguish Iyl from apical and dorsal consonants, I have included it in the lamino-palatal column in the consonant chart.)

Initial /y/ before vowel /i/ or /a/ may be dropped (that is the vowel may be pronounced without the initial onglide) and younger speakers, up to about 20 years of age, have even been recorded as substituting glottal stop for Iyl in this position. Thus there are three alternative pronunciations for words beginning with /y/ followed by /i/ or /a/, for example:

yinya, inya, ?inya 'that' yanyu, anyu, ?anyu 'this'

Semivowel /y/ after /i/ word-finally or before another consonant is always deleted (see §2.5.2).

Initial /w/ before /a/ is clearly pronounced (there are no examples of initial /w/ before /i/), whereas initial Iwl before lui may be dropped, giving the two alternatives:

wunba, unba 'English bee',

but no glottal stop has been recorded in this position in place of Iw/.

2. 1.2 Vowels

Kuku Yalanji has the common Australian inventory of three vowels: /a/, /u/ and /i/. As is characteristic for Australian vowels, they have 'pure' articulation and are never diphthongised in any way. There is no phonemic distinction of vowel length as in the neighbouring languages Dyabugay and Guugu Yimidhirr. Generally, vowels in Kuku Yalanji could be classified as short, with slightly longer allophones occurring in word-final position.

While some Australian languages with the same vowel system allow each vowel a fairly wide range of phonetic realisation (see Dixon 1 980: 1 30), Kuku Yalanji vowels show only little variation in different environments. Figure 2.1 shows the range of vowel realisation in Kuku Yalanji with * marking the approximate place of articulation of the major allophone.2

The major allophone of vowel /a/ is more central and slightly higher than cardinal vowel /a/. Before velar consonants, lateral /l/, retroflex rhotic /r/ and semivowel /w/ it is pronounced a little further back, whereas before laminal consonants and semivowel /y/ it is slightly fronted and raised. The major allophone of vowel /i/ is less high and less front than cardinal vowel /i/ and is pronounced with spread lips. Before velar consonants, bilabial nasal /m/,

2 In my description of Kuku Yalanji vowels I follow the traditional vowel classification which is based on a correlation of highest point of the tongue and shaping of the lips as represented in stylised form in the vowel quadrilateral (O'Connor 1 973; Ladefoged 1 975; Catford 1 977).

22 Chapter 2

lateral /l/ and retroflex rhotic /r/ it is lowered somewhat and pronounced in a more central position with less lip spreading. The major allophone of vowel /u/ is more fronted and lowered than cardinal vowel /u/ and is pronounced with distinct lip rounding. It tends to be lowered in the same environment where /i/ is lowered, and is more fronted before palatal consonants, but distinct lip rounding applies throughout.

Figure 2.1: Range of phonetic realisations of Kuku Yalanji vowels

2.2 Phonotactics

The majority of Kuku Yalanji roots are disyllabic. Except for three monosyllabic interjections there are no monosyllabic roots. In a word list of 1 340 roots3 there are 1 055 disyllabic and 200 trisyllabic roots, 81 roots with four syllables, and 4 roots with five syllables. All roots begin with a consonant and end in a consonant or a vowel. Thus the structure of Kuku Yalanji roots can be formalised as follows:

$C_1 (V C_2)^n V (C_3)$ where $n \ge 1$

2.2. 1 Possibility of occurrence

The following vowels and consonants can occur in the different phonotactic positions in Kuku Yalanji root structure:

• V can be any vowel.
• $C_1$ must be a single consonant and can be any stop, nasal or semivowel, i.e. /b, d, j, k, m, n, ny, ng, w, yl (see Figure 2.2). One word with initial rhotic IITI has been recorded, rrunyuji 'wise man with magic powers'. This word could perhaps have been part of a ceremonial speech style for which there is no more evidence today. Phonological patterns that are different from the everyday language are not unheard

³ This word list is based on two sources: the large majority of entries were elicited by myself and a smaller number of lexical items is taken from the word list by W. Oates and L. Oates ( 1964b) and an unpublished preliminary word list by H. Hershberger and R. Hershberger (early 1 960s). Items from these lists were only included in my data after thorough checking with informants. The dictionary published by the Hershbergers in 1 982 has not been taken into account for this study.

of for such speech styles in Australian languages, although they are uncommon (see Dixon 1 980:60).

• C3 must be a single consonant and can be a non-peripheral nasal, the lateral, either rhotic or the lamino-palatal semi-vowel, i.e. In, ny, 1, rr, r, yl (see Figure 2.2).

Figure 2.2: Initial and final consonants in Kuku Yalanji

• C2 can be:

(i) any single consonant, or a cluster of two or three consonants.

These clusters may be:

• (ii) a homorganic nasal/stop sequence, i.e. 1mb, nd, nyj, ngk/;
• (iii) apico-alveolar nasal Inl plus: a non-apical stop, or a peripheral nasal, i.e. Inb, nj, nk, nm, nng/;
• (iv) semi-vowel Iyl plus: a non-apical stop, or a bilabial nasal, or a peripheral homorganic nasal/stop sequence, i.e. Iyb, yj, yk, ym, ymb, yngk/;
• (v) lateral /l/ plus: a non-apical stop, or a peripheral nasal, or a homorganic nonapical nasal/stop sequence, i.e. lIb, lj, 1m, lng, 1mb, lnyj, lngk/;
• (vi) trilled rhotic Irrl plus: a peripheral stop, or a peripheral nasal, or a homorganic peripheral nasal/stop sequence, i.e. Irrb, rrk, rrm, rrng, rrmb, rrngk/;
• (vii) retroflex rho tic Irl plus: a peripheral stop, or a non-palatal nasal, or a homorganic non-palatal nasal/stop sequence, i.e. Irb, rk, rm, rn, rng, rmb, rnd, rngk/. Clusters Irnl and Irndl stand out from the general pattern of consonant clusters in Kuku Yalanji (see Table 2.2) and there are only one and two occurrences respectively which are:

murnil	'stir, twist'
warndiy	'get up from sitting or lying position'
kurndal	'sink in mud or water'.

nb	nj	nk	nm	nng
yb	yj	yk	ym	?	ymb	?	yngk
lb	lj	lk	lm	lng	lmb	lnj	lngk
rrb	(?)	rrk	rrm	rrng	rrmb	(?)	rrngk
rb	(?)	rk	rm	rn	rng	rmb	rnd	(?)	rngk

Table 2.2: Heterorganic consonant clusters in Kuku Yalanji

Cluster IrndJ has been attested once in Dyabugay, (barndil 'child'), but no example of 1m! is known for Dyabugay. Neither cluster exists in Yidiny and Dyirbal, further south (Dixon 1 977:35 and 1 972:272-273 respectively), but they are present among the possibilities for consonant clusters in Guugu Yimidhirr to the north of Kuku Yalanji. However, these clusters do have some idiosyncracies in Guugu Yimidhirr in that Irnl may be pronounced as either a cluster of distinct sounds or as a single retroflex nasal, and Irndl as either a sequence of rhotic plus apical nasal and stop, or as retroflex nasal followed by retroflex stop, apparently in free variation (see Haviland 1 979a:37, 40; Dixon 1 980: 1 64). Apparently there are no cognates in Guugu Yimidhirr for the Kuku Yalanji words containing these clusters (Haviland, unpublished word list February 1 981) which makes borrowing from Guugu Yimidhirr into Kuku Yalanji appear unlikely. A more plausible assumption would be that these clusters are an area feature, albeit weak, in the three languages north of Cairns.

Clusters Iyngl and Iynyjl (marked '?') for which no examples have been attested, could reasonably be expected to exist in Kuku Yalanji according to the general pattern, and their absence here may just be accidental gaps in a relatively small word list of only 1 340 roots. The gaps marked by '(?)', on the other hand, are so symmetrical as to suggest the general rule that a rhotic may not be followed by a palatal consonant in Kuku Yalanji. This would represent a special feature of Kuku Yalanji, since northern and southern neighbouring languages do allow such sequences.

Possibilities for intermorphemic consonant clusters are less restricted than those for intra morphemic clusters. Thus in reduplicated forms, compounds, and some inflected forms we can find clusters across the root boundary that cannot occur within the root, I.e. combinations of C3 and Cl. For instance:

reduplicated forms:

walbulwalbul	'butterfly'
dakaldakal	'trochus shell'
nurilnuril	'keep looking inside'

compounds:

miyil-dandi	'sleepless' (lit. eye-hard)
julbarr-warriy	'slip' (lit. slippery-run)

Compounds are defined as such because they inflect as a whole, have primary stress on the first syllable (see §2.6.1 ), and no other morpheme such as a clitic or particle can be inserted between the two parts. Therefore the above examples can be regarded as one word, grammatical as well as phonological, although I hyphenise analysable compounds for clarity.

Inflected nominal forms showing further clusters are:

dambalda	'boot-INST'
manyarrda	'wife-ERG'

On the other hand, most inflected forms avoid any kind of consonant cluster across the morpheme boundary by inserting a vowel between a root-final consonant and the inflection, for example:

bama-ngka	'Aborigine-ERG'
maral-angka	'girl-ERG'
bama-nda	'Aborigine-LOC'
maral-anda	'girl-LOC'

One restriction on intermorphernic consonant clusters is that Kuku Yalanji does not allow geminate consonants, so that a cluster of two identical consonants is reduced to one consonant. For instance, in the reduplication of y-conjugation verbs consonant /n/ is inserted between the roots (see §3.8.5.5 (ii) (a)). If the root begins with /n/, the resulting cluster /nn/ is reduced to single /n/, for example:

nuka-n-nuka-ji-y ‘keep eating (itr)’

→ nuka-nuka-ji-y.

Bound morphemes, that is inflections, derivations and clitics, differ in their structure from roots in that they are frequently monosyllabic and may even consist of a single consonant or vowel. Disyllabic bound morphemes conform to the phonotactic occurrence possibilities of roots, whereas monosyllabic bound morphemes show some variation: they may begin with a consonant cluster, e.g. potent ergative -ngkV, potent locative -ndV, and there is one occurrence of morpheme-initial lateral in clitic -lu/-la. (The presence of initial vowel in a bound morpheme, something which is not possible for a root, is conditioned by root-final consonant, e.g. potent ergative -VngkV/C-, and should therefore not be regarded as a special feature of bound morphemes that goes against the general phonotactic conventions.)

2.2.2 Probability of occurrence

The various phonemes of Kuku Yalanji have different relative probability of occurrence, partly depending on the presence of other phonemes within the same word. Occurrence rates for Kuku Yalanji phonemes are based on a lexicon of 1 340 roots (excluding the phonetically unusual form rrunyuji 'wise man with magic powers').

The relative frequency of the three vowels in Kuku Yalanji is listed in Table 2.3. The percentages in this table show that overall /a/ is more frequent than /u/, which in turn is more frequent than /i/. The relatively higher occurrence of /i/ over /u/ in third and fourth vowel positions can be wholly accounted for by the fact that a number of nominal roots with more than two syllables appear to be fossilised comitative forms ending in -ji, and several verb roots of more than two syllables are reinterpreted roots ending in -rri (see §3.8.3.2 (ii)).

	a	u	i
1 st syllable	46.6%	35.2%	18.2%
2nd syllable	54.2%	27.1%	18.7%
3rd syllable	42.3%	19.8%	37.9%
4th syllable	61.1%	18.9%	20.0%

Table 2.3: Relative frequency of Kuku Yalanji vowels

Looking at the distribution of vowels in first and second vowel position of a root, we find that vowels follow each other with either higher or lower frequency than could be expected. For instance, 35.2% of roots have vowel /u/ in first position and 27.1% have /u/ in second position. If there was no association between vowels in first and second position, we would expect $0.352 \times 0.271 = 0.095$ , i.e. 9.5% or 128 roots to have /u/ in both first and second vowel position. In fact, there are 200 roots, or 14.9%, with /u/ as first and second vowel. Thus the actual-to-expected ratio of /u/ following /u/ is $0.149 \pm 0.095 = 1.57$ . In other words, /u/ follows /u/ more often than expected, and we can say that vowel /u/ has positive association with following vowel lui. The actual-to-expected ratio for each vowel cooccurrence is listed in Table 2.4. These figures show that each vowel in Kuku Yalanji has highest association with the same vowel and relatively lesser association with any other vowel, as set out in Table 2.5. Note that the markedly highest association is between /u/ and /u/, whereas /i/ has some positive association with /a/, although less than with /i/, but overall the least association with /u/.

Table 2.4: Actual-to-expected ratio of vowel co-occurrence in first and second vowel position

	Second position
	a	u	i
First position	a	1.13	0.73	0.93
	u	0.79	1.57	0.79
	i	1.04	0.59	1.47

Table 2.5: Relative association between vowels in Kuku Yalanji

	greatest association with	medium association with	least association with
a	a	i	u
u	u		a, i
i	i	a	u

This pattern is interesting with regard to the vowel-harmony rule for suffixes (see §2.S. I) which requires that /a/ and /i/ be followed by /a/, and /u/ be followed by /u/. It appears that this vowel-harmony rule reflects to some extent the general tendency of vowel co-occurrence within roots.4

Occurrence rates for initial and final consonants are given in Table 2.6. Final consonants are counted only for non-verbal roots, since all verb roots end in a vowel with the final consonants -y and -l in citation form being conjugation markers. Adjectives and nouns each show a slightly higher rate of consonant-final than of vowel-final roots; that is, there is no structural difference between adjectives and nouns concerning the final root constituent. Of a total of 1066 noun and adjective roots, 53% end in a consonant and 47% in a vowel.

It can be seen that more than half of all roots begin with a stop. The most frequent initial segment is k and the least frequent n. Of the 582 roots that end in a consonant more than half show a liquid in final position. Lateral /l/ is overall the most favoured consonant in final position. According to place of articulation, the most frequent consonants root-initially are labials, followed by palatals and velars with equal occurrence rates, the least frequent consonants are alveolars. The latter, however, have the highest occurrence rate root-finally (owing to the high frequency of liquids). Overall the least common initial consonants are non-peripheral nasals, with palatal nasal Inyl also being quite rare root-finally.

	root initial					root final
	lab.	alv.	pal.	vel.		(post-) alv.	pal.
b	16%				} 58%
d		9%
j			15%
k				18%
m	11%				} 22%
n		1%				19%
ny			2%				3%	} 22%
ng				8%
l					} 62%	33%
rr						18%
r						11%
w	11%				} 20%
y			9%				16%
	38%	10%	26%	26%		81%	19%

Table 2.6: Frequency rates of initial and final consonants by place and manner of articulation

4 A similar pattern of weak vowel association with roots can be found in Guugu Yimidhirr (Haviland 1979a:39), which also has a vowel-harmony rule for imperative suffixes on -rr conjugation verbs.

The distribution of single consonants and consonant clusters in C2 position shows the following pattern: Liquids and stops $/j/$ and $/k/$ occur with highest frequency. Bilabial stop $/b/$ is the next most common consonant in medial position, followed in frequency by semivowels, bilabial nasal $/m/$ , apical stop and nasal $/d/$ and $/n/$ , and velar nasal $/\text{ng}/$ in this order. The least frequent single consonant in C2 is laminal nasal $/\text{ny}/$ . Among 574 occurrences of consonant clusters, homorganic nasal/stop sequences are the most common ones with an occurrence rate of 34%. Clusters with the lateral or rhotic Irrl as first component followed by a peripheral stop are also fairly frequent (26%). All other clusters have an occurrence rate of less than 5%, the least common being Inm! and Inngl (twice each), Irnl (once) and Irndl (twice), and Iym! (twice), (ranging between 0. 17 and 0.34%).

On the whole there is no clear association in Kuku Yalanji between a consonant and a following or preceding vowel (unlike in Dyirbal; see Dixon 1 972:28 1 -283). The actual occurrence rates for CV and VC combinations generally follow the pattern of expected occurrence rates with only a very slight tendency toward positive association between consonants and vowels that are close in place of articulation and negative association between consonants and vowels that are more distant in articulation. This tendency is most obvious for velar consonants and semivowels. Initial velar stop and nasal $/k/$ and $/\text{ng}/$ and semivowel $/w/$ are markedly less often followed by front vowel $/i/$ than could be expected, whereas semivowel $/y/$ is markedly less often followed by back round vowel $/u/$ . In these cases the consonant clearly has least association with the vowel that is most distant in articulation. VC sequences $/\text{ing}/$ and $/iy/$ , on the other hand, show a surprising departure from this general pattern. Sequence lingl occurs with 1.1 % frequency, whereas only 0.8% would have been expected, and liyl shows less than expected frequency (0.5% versus expected 1.6%). The latter discrepancy may be accounted for by the fact that $/y/$ after $/i/$ word-finally or before another consonant is lost (see §2.5.2). But there is no apparent clue to the reason for the relatively high occurrence of sequence ling!.

2.3 Phonological adaptation of English loan words5

When the Kuku Yalanji people came into contact with European civilisation they used two strategies for naming newly introduced commodities. One way was to extend the meaning of Kuku Yalanji words to describe these novelties, as for instance, kambi 'flying fox' for 'clothes' (the flying fox wraps its wings around itself when at rest), diburr 'eggs' for 'bullets', mayi karraji 'fruit with strings' for 'mangoes' (mangoes hang off the tree on long string-like stalks), orjiri-burr 'through the sky' for 'aeroplane'.

Another way of naming things was to use their English name, appropriately adapted to Kuku Yalanji phonology. This process of phonological adaptation, however, seems to have ceased about one generation ago (see Donaldson 1 980:62), presumably with the increase of Kuku Yalanji I English bilingualism. Nowadays one frequently hears words like 'car', 'aeroplane', 'church', 'store', or 'soft-drink' in the Kuku Yalanji discourse of younger speakers, usually with appropriate inflectional affixes (see §3. 1 2), but without phonological adaptation. This new trend also includes the frequent borrowing of English verbs and adjectives, which apparently was not commonly done in the earlier days, again with

⁵ For a detailed discussion of this topic see Dixon, Ramson and Thomas ( 1990), Chapter 5 'Borrowings into Australian languages'.

l

morphological, but not phonological adaptation, such as fix-im-bungal, shift-mal-mal, workmani-jiy, or blue-bala, dirty-bala (see §3.12).

Only older speakers (roughly over 60 years of age), who are less proficient in English, still modify some newer words to some extent (usually not totally), e.g. telelzision, or job job for the community's favourite television program in 1 980 Cop Shop. The vocabulary of such older speakers also contains phonologically well-adapted loan words from the earlier contact period (which usually have English pronunciation among the younger speakers), e.g. bulawu 'flour', banabul 'pineapple', mudaka 'motorcar', juka 'sugar'.

Only about a score of English loans have become an integral part of the Kuku Yalanji lexicon. From these, and from the few adaptations still made by older speakers, we can glean the following nine generalisations on phonological change in English loan words (see Dixon 1 972:326):

(i) Kuku Yalanji stops b and j correspond to English fricatives that are closest in place of articulation, i.e. /b/ corresponds to /v, f/ and /j/ to /s, z, ʃ, ʒ/, for example:

naybu	'knife'
aybi	'Ivy (name)'
buji	'pussy (cat)'
juka	'sugar'
maji	'matches'

An interesting loan which cannot be placed with any amount of certainty is binjin 'petrol'. I have not been able to ascertain an early use of English benzene for petrol, and it might be possible that this is a loan from German, Benzin 'petrol', introduced by the German missionaries.

• (ii) English voiceless and aspirated stops become unaspirated and usually voiced, e.g. dubayku 'tobacco', muduka 'motorcar'. (Note that k in Kuku Yalanji normally represents voiced [g].) The characteristic acoustic features of changes (i) and (ii) were aptly described by my informant Bobby Roberts, who said: "English people say 'pussycat'. Bama say that word too, but we say it lightly way, we don't say phussi or phussikhath, but buji or bujiga".
• (iii) English It I and lsi before a velar stop become Kuku Yalanji IITI, as in marrkin 'musket (rifle)' and warrkin '(a place name) from 'Watkins' (name of an English settler)'.
• (iv) English consonant clusters consisting of a stop or fricative plus a sonorant are broken up by insertion of a vowel, for example:

jarruja	'trousers'
gilaja	'glass (mirror)'
bulawu	'flour'
banabul	'pineapple'

If an initial cluster is broken up in this way, the inserted vowel is stressed according to the word-stress rule (see §2.6. 1 ), e.g. jarruja. (Examples are too few to determine rules for the choice of vowel that is inserted.) An initial fricativelstop consonant cluster, on the other hand, is simplified by dropping the first consonant, e.g. buwun 'spoon' (which is the only example in my data).

• 30 Chapter 2
• (v) English loans with initial vowel are preceded by a semivowel which is Iwl before a back vowel and Iyl before a front or central vowel, for example:

wulman	'old man'
yalibala	'early'

• (vi) An initial consonant which is not permitted in Kuku Yalanji phonotactics, but which is retained in loan words, is lateral /l/, e.g. landin 'Landing' (name for the confluence of Thompson Creek and Bloomfield River where there used to be a landing ramp). No loans with original initial rhotic have been attested.
• (vii) Monosyllabic English roots are made disyllabic by either inserting a semivowel sequence within the root:

buwun	'spoon'
biyin	'(safety) pin',

or by adding a final vowel, if the original or adapted root ends in a stop, for example:

biki (or bikibiki)	'pig'
juki	'chook, (hen)'
baybu	'pipe'
naybu	'knife'

(Again, no generalisation on the choice of final vowel is possible.)

• (viii) Final consonants that are not possible in Kuku Yalanji phonotactics are adapted in the following way:
  • (a) by adding a vowel after an original or adapted stop as in (vii) and also gilaji 'glass', mijiji 'Mrs';
  • (b) by changing a stop to a corresponding nasal, as in marrkin 'musket';
  • (c) by changing a peripheral nasal to a non-peripheral one, as in landin 'Landing';
  • (d) or by simply dropping the final consonant, as injarruja 'trousers', bujika 'pussycat', warrkin 'Watkins'.
• (ix) The examples in my data do not give sufficient evidence for a generalisation on the adaptation of vowels, except to say that Kuku Yalanji vowels on the whole approximate the English vowel, and that English [E] or [re] and diphthong [ail before a consonant tend to become layl in Kuku Yalanji, e.g. mayngku 'mango', dubayka 'tobacco', baybu 'pipe'.

2.4 Phonetic rules - assimilation

There are three regressive assimilation rules in Kuku Yalanji. These are:

There are three regressive assim

(i)

$$a, u \rightarrow a^{i}, u^{i} / - \begin{Bmatrix} ny \ j \end{Bmatrix}$$

(ii) $n \rightarrow n_j$ / -j (iii) $k \rightarrow k^w$ / -u

These rules apply with varying degrees of consistency among speakers.

According to rule (i), vowels /a/ and /u/ have a palatal offglide if they precede a palatal nasal or stop, for example:

nukainy	'ate'
jalbainy	'taboo food'
wunkuiny	'next day'
jukumuiny	'from the tree'
waijul	'burn, cook'
nguijan	'jealous'

Generally, palatalisation is stronger with vowel /a/ than with vowel /u/.

Rule (ii) determines that apical nasal /n/ is palatalised before palatal stop /j/, for example:

Yalanjji	'(language name)'
baykanjji	'lest (it) bite'.

This rule would seem to negate the phonemic distinction between clusters /nj/ and /nyj/ in Kuku Yalanji. However, assimilation rule (i) helps to show the difference between the two clusters. A palatalised nasal before /j/ does not in turn influence its preceding vowel, whereas a phonemic palatal nasal does; that is, assimilation rule (i) precedes assimilation rule (ii). Consider the following minimal pairs:

binainyji	'fig-like fruit'
binanj-ji	'with a rudder'
jalbanj-ji	'with a tree top'
jalbainy-ji	'with taboo food'

According to assimilation rule (iii), velar stop /k/ is labialised before round vowel /u/, as for instance in:

dakwuy	'hungry'
nyikwu	'now, today'.

This rule shows greatest variation between speakers, some of them clearly labialising the velar stop and others not at all.

2.5 Morphophonological processes

2.5. 1 Vowel harmony

For a number of suffixes in Kuku Yalanji the suffix vowel is selected according to vowelharmony rules relating to the last root- or stem-vowel. The affected suffixes are: all allomorphs of nominal inflections ergative, dative, locative, and perlative (with two notable exceptions to be discussed below), verbal subordination suffix -nyV and 'first of all' clitic -ngVrr. For these suffixes the following vowel harmony rules apply:

$$\begin{array}{rcl} \text{suffix vowel} = & u & / & u(\mathbf{C}) \text{-} \ & a & / & \left{ \begin{array}{c} a(\mathbf{C}) \ i(\mathbf{C}) \end{array} \right} \end{array}$$

32 Chapter 2

In the case of u/u- and a/a- there is clearly total assimilation to the preceding vowel; however, this does not apply where the last stem vowel is /i/.

Data from Wagaman (Dixon, fieldnotes; see also Dixon 1 980:178- 1 79), an inland dialect of the same language as Kuku Yalanji, may give some indication on how this partial vowel harmony developed. Table 2.7 shows the vowel distribution in ergative, dative and locative inflections in Wagaman.

	a-	u-	i-
Ergative	- $\eta$ ga (one occurrence of - $\eta$ gu )	- $\eta$ gu	- $\eta$ gu
Dative	- ga (sometimes - gu )	- gu	- gu
Locative	- $\eta$ a	- $\eta$ u (sometimes - $\eta$ a )	- $\eta$ a

Table 2.7: Vowel harmony in Wagaman inflections

It appears that these case inflections in Wagaman have the underlying forms: ERG - $ŋ$ gu, DAT -gu, LOC -lJa, which correspond to case inflections in languages that have no vowel alteration. Then the following vowel-harmony rule applies:

suffix vowel	$u \rightarrow a/a(C)-$
suffix vowel	$a \rightarrow u/u(C)-$

but after final stem vowel /i/ - the suffix vowel remains unchanged.

It seems that in Kuku Yalanji the harmony pattern for the underlying a-final locative suffix has been generalised to apply to u-final suffixes such as ergative and dative as well, and has then also spread to other suffixes. For dative and locative the vowel-harmony rules apply totally and no underlying canonical forms -ku DAT or -nga LOC can be determined. Ergative and perlative, on the other hand, each have one allomorph which is never affected by vowel harmony. This is neutral ergative -bu/V and perlative -burr/V which could be regarded as the canonical forms. (Note the typical vowel luI in the ergative inflection.)

It seems reasonable to assume that Kuku Yalanji vowel harmony has developed from the harmony rules that apply in Wagaman, since the present Kuku Yalanji system is more comprehensive and general than the Wagaman rules. The generalisation of the harmony pattern for locative inflection (as in Wagaman) could be explained in the following way. Locative in Kuku Yalanji has a large variety of functions (see §4. 1 .4.4 (iii)), including some references to goal, which in other languages is expressed by dative. Locative therefore has a higher occurrence rate than dative in Kuku Yalanji. If we look at the relative distribution of locative and ergative in discourse, we find that an agent NP is very often a pronoun, where ergative does not apply. Locative inflection, on the other hand, is the same for nouns and pronouns, so that again locative inflections have a relatively higher occurrence rate than ergative inflections. Thus it would appear that Kuku Yalanji vowel harmony has been generalised following the pattern of the most frequently occurring inflection. Furthermore, the fact that the last root vowel is much more often /a/ than /u/ or /i/ (see §2.2.2, Table 2.3) could have influenced the vowel-harmony rule in Kuku Yalanji. If the Wagaman system applied at some stage in Kuku Yalanji, inflectional suffixes would have contained vowel lal much more frequently than /u/ or /i/. Again, the most common form could then have been

generalised to apply also after root vowel /i/, which occurs least frequently, whereas root vowel /u/ maintained its influence on the following suffix vowel. This also corresponds to the pattern of vowel association that is found within roots, i.e. that /a/ is followed more often by $/a/$ than by any other vowel and $/i/$ has at least some positive association with $/a/$ , whereas $/u/$ has a strong tendency to be followed by $/u/$ (see §2.2.2, Tables 2.4 and 2.5).

2.5.2 Deletion of semivowel /y/

After vowel $/i/$ in word-final position or before another consonant, semivowel $/y/$ is not pronounced. Thus we have the rule:

$$\mathbf{y} \stackrel{\ast}{\to} \mathfrak{g} \quad \prime \stackrel{\ast}{\to} \mathbf{y} \quad \prime \begin{Bmatrix} \mathfrak{g} \ \mathbf{C} \end{Bmatrix}$$

Because of this deletion rule, roots ending in -iy cannot be distinguished from those ending in -i in citation form, but they are distinct in some inflected forms. The neutral locative inflection, for instance, is -ng $V/V$ - and -mb $V/y$ - (with different allomorphs after other consonants). Consider the following roots in citation form (absolutive) and with locative inflection:

Absolutive:	kulji	'stone'
	dili	'corkwood pine'
	duli	'burnt grass'
Locative:	kulji-nga	'on/to the stone'
	dili-mba	'on/to the corkwood pine'
	duli-mba	'on/to the burnt grass'

From the inflections we can infer that dili and duli have underlying forms with final $/y/$ , diliy, duliy, whereas kulji is a genuinely i- final root. These examples also illustrate the loss of $/y/$ after $/i/$ before a following consonant. Within a root, $/y/$ cannot be recovered in this environment, but the surprisingly low occurrence rate of $/iy/$ in medial position (see §2.2.2) may indicate that some instances of underlying Iyl have been lost.

In order to indicate which inflectional allomorph is used with a particular nominal root, I represent such roots in citation form with final -i or -iy respectively. (Here I differ from the Hershberger spelling system which does not represent final y after i-.)

Similarly, verbal nonpast inflection -y is not pronounced after roots or stems ending in -i. However, rather than introducing an additional nonpast inflectional allomorph of - $\phi$ / i-, I represent all y-conjugation verbs with final -y in nonpast tense and in citation form (-y is also the conjugation marker), for example:

citation forrnlnonpast:

badiy	'cry, weep'	pronounced:	badi
karangajiy	'sneak up'		karangaji
manijiy	'get (itr)'		maniji

analogous to: bunday 'sit' janay 'stand'

2.5.3 Syllable reduction

The following three syllable reductions have been observed in normal speech:

(i) Locative inflection -nga after root-final /a/ is reduced to root-final long vowel a:, that IS:

Locative -anga � a:

as for instance bana: 'inlto the water' instead of bananga. The reduction of locative inflection to long vowel a: has not been heard more than perhaps half a dozen times in narratives and has always been corrected to -nga by the informant when playing back the tape. But even if it is used only occasionally and is not accepted as an alternative to the full inflection, this reduced form is very interesting when considering Djabugay. In Djabugay the locative inflection for roots ending in a vowel is lengthening of the final vowel (Patz 1 99 1 :264) and the same applies for Yidiny on even-syllabled stems (Dixon 1 977: 1 28). This supports the previous suggestion (see §2.1.1) that Kuku Yalanji is the more conservative and Dyabugay the more innovative of the two languages.

• (ii) The causative verb-forming derivation -bungal is frequently reduced to -bal, particularly in reduplicated forms, e.g. kima-bungal or kima-bal 'make something soft', dandi-bungal-bungal or dandi-bal-bal 'keep making something hard'. Here, not all speakers correct themselves when they have used the reduced form, maintaining that either form is fine. It appears that the reduced form -bal may be in the process of becoming the accepted norm.
• (iii) In some reduplicated verbs a reduced form is already firmly established as the norm. Usually, verb roots reduplicate as a whole, but six very commonly used verbs show only partial reduplication of either the first or the second syllable and these forms are always used instead of the wholly reduplicated forms which by now are even unacceptable. These verbs are:

Reduplication of first syllable:

banbadiy	*badi-n-bady	'keep crying'
janjanay	*jana-n-janay	'keep standing'
kankaday	*kada-n-kaday	'keep coming'
wanarriy	?warri-n-warriy	'keep running'
Reduplication of second syllable:
bundanday	*bunda-n-bunday	'keep sitting, stay'
wunanay	*wuna-n-wunay	'keep lying, sleep'

(For a full discussion of form and function of verb reduplication see §3.8.S.S (ii).)

Another partially reduplicated verb form that is used occasionally is nukalkal instead of nuka-I-nuka-l 'keep eating'. Again, only the second syllable is reduplicated in the reduced

form. So there are four examples of first-syllable reduplication and three examples of second-syllable reduplication, and the question arises why it should sometimes be only the first and sometimes only the second syllable that is reduplicated. Second-syllable reduplicated forms share the common aspect that the syllable containing the most frequent vowel /a/ is reduplicated and the syllable containing a less frequent vowel is dropped. The same pattern is evident in the reduction of -bungal to -bal, and also in the forms banbadiy and wanarriy with first-syllable reduplication. One could therefore make the tentative suggestion that in a reduction process the syllable with the less frequent vowels /u/ and /i/ is deleted and the syllable containing the most frequent vowel /a/ is maintained.

2.6 Stress and intonation

Stress, which is characterised by higher pitch and loudness (see Ladefoged 1 975:222ff.), identifies a word as a unit in Kuku Yalanji. Clausal intonation contours, the highest point of which is the clause stress, show some variation and are the only distinguishing feature for certain types of clauses. H. Hershberger and E.v. Pike (1 970) have done a detailed study of stress and intonation in Kuku Yalanji. Unless stated otherwise, my own findings concur with Hershberger and Pike, but I have generalised some of their rules and addressed some further aspects.

2.6. 1 Word stress

Every word in Kuku Yalanji has primary stress on the first syllable, after which there is a distinct drop in pitch for disyllabic words and a more gradual drop for words of more than two syllables (see Hershberger & Pike 1970:791-792). Words with four or more syllables that are roots, reduplicated forms, or compounds, may have secondary stress on the third syllable, that is the first syllable of the second root in reduplicated forms of compounds. A reduplicated trisyllabic root may have secondary stress on the fourth syllable, again the first syllable of the second root. (There are apparently no compounds where the first constituent has more than two syllables.) Thus we can set up the general word stress rule:

$V \rightarrow + \text{stress} / # \text{root} [C (VC_1VC_1)] -$

where C1 may be a single consonant or a cluster. For example: (primary stress, secondary stress)

jálbu	'woman'
bájibay	'bone'
wáwubàja	'river'
yálmburràjaka	'whale'
kángkal-kàngkal	'own children'
mílka-wùlay	'forget' (lit. ear-die)
míyil-dùdu	'illiterate' (lit. eye-blunt)
búlkiji-bùlkiji	'pipis'

36 Chapter 2

However, secondary stress is not obligatory and is usually omitted in other than careful citation forms, leaving the word with stress only on the first syllable.

Derived stems (other than compounds or reduplicated forms) and inflected forms with or without clitics, for which no detailed stress patterns are indicated by Hershberger and Pike, receive secondary stress as follows:

(i) Disyllabic inflections or derivational suffixes are treated like roots and may have secondary word stress on the first syllable, for example:

máral-àngka	'girl -ERG:pt'
ngánjan-ànda	'father -LOC:pt'
bímakay-kàrra	'FZ-KPL'

(ii) Monosyllabic inflections, derivational suffixes or clitics are unstressed. But if the resulting word has more than three syllables, the penultimate syllable receives secondary stress, unless it is the last root syllable, in which case it remains unstressed, for example:

báma-ngka	'Aborigine-ERG:pt'
báma-ngkà-ku	'Aborigine-ERG:pt-EMPH1'
báma-ngka-kù-lu	'Aborigine-ERG:pt-EMPH1-EMPH2'
búlkiji-ku	'pipi-EMPH1'
búlkiji-mùn-bu	'pipi-mun-LOC'6
búlkiji-mun-bù-ku	'pipi-mun-LOC-EMPH1'
yárraman-ànda-kù-lu	'horse-LOC:pt-EMPH1-EMPH2'
bímakay-kàrra-ngka-kù-lu	'FZ-KPL-ERG:pt-EMPH1-EMPH2'

However, this secondary stress rule does not apply to the combination of suffix + clitic -baja + ku 'very' attached to an adjective (see §3.2.3 .8). None of the syllables in this sequence are stressed, e.g. yalbay-baja-ku. (See Hershberger and Pike 1 970:802, who regard bajaku as a 'word'; however, the consistent absence of stress contradicts this interpretation. )

But again, secondary stress is not obligatory and is in fact apparent only in slow and careful speech. In normal speech, even in long words of up to eight syllables, only the first syllable is stressed, with pitch subsiding over the following two or three syllables and remaining level over the rest. (Such long words are rare, but stems with inflections and clitics consisting of up to five or six syllables are fairly common in discourse.)

2.6.2 Clause stress

Clause stress conforms to the word-stress pattern; that is, it falls on the first syllable of a word. It is characterised by higher pitch and loudness as compared to other clause constituents where regular word stress is less prominent.

Hershberger and Pike ( 1 970:792-802) posit rules for obligatory and optional stress for different types of declarative and interrogative clauses. I suggest that it is possible to generalise these rules, saying that each clause has an unmarked stress pattern which may be

⁶ For Catalytic suffix -mun- see §3.2.3. 1.

changed, if the speaker wants to assign special emphasis to a particular clause constituent. The unmarked stress patterns are as follows:

• (i) In a simple statement with only two constituents, subject (topic) and predicate (comment), clause stress, indicated by , before the vowel, falls on the predicate (see Hershberger & Pike 1970:794), for example:
  • (1) lawunkarra k'aday. '(Our) friends are coming.'
  • (2) Kirrbaji y'albay. `The dugong is big.'
• (ii) In a simple statement with more than two constituents, clause stress falls on the constituent immediately preceding the predicator (see Hershberger & Pike 1970:793, 3.1), for example:
  • (3) Kayangka b'ajibay nukany. `The dog ate a bone.'
  • (4) Kaya m'inyaka wawu. 'The dog wants meat.'
  • (5) Kayangka ngayku m'inya nukany. ‘The dog ate my meat.’
  • (6) Kayangka yunu minya ng'awuya nukany. 'The dog ate your turtle meat.'
  • (7) Kaya j'inbal wanarriy. ‘The dog is running fast.’
  • (8) Nyulu y'ilayku kadany. ‘He came yesterday.’
  • (9) Maral b'ayanba bundanday. ‘The girl is staying in the house.’

Hershberger and Pike (1970:801) state that if the stressed clause component is a noun phrase with two or more constituents, each of these constituents is stressed equally strongly. However, according to my observations, it is only the last word in the phrase which receives most prominent clause stress, as in (5) and (6).

The same unmarked pattern of pre-predicator clause stress applies for negative statements; that is, stress is on the negative particle kari, which precedes the predicator:

• (10) Nyulu jarba k'ari nyajiny. 'He did not see the snake.'
• (11) Ngayu mayi diburr k'ari nukany. ‘I did not eat the egg(s).’
• (iii) (a) In a content interrogative clause the interrogative pronoun is stressed (see Hershberger & Pike 1970:195):
• (12) W'anyangka mayi nukany? 'Who ate the food?'
• ( 1 3) Yundu w'anyu nyajiny? 'What did you see?'
• (14) Yundu w'anjabu dungay? 'Where are you going?'

The above sentences show unmarked word order, but an interrogative pronoun may occur anywhere in the clause and still attracts clause stress, for example:

• ( 1 2) a. Mayi w'anyangka nukany?
• (13) a. W'anyu yundu nyajiny?
• (14) a. Yundu dungay w'anjabu?
• (iii) (b) In a polar question the predicator is stressed (see Hershberger & Pike 1 970:796):
  • ( 1 5) Yundu jalbu ny'ajiny? 'Did you see the woman?'
  • (16) Yundu d'akuy? 'Are you hungry?'
  • (17) Nyulu d'ungany? 'Did he go?'

This means that polar questions and statements with only two constituents have basically same stress assignment (see (i) above); however, they differ in the level of pitch, which will be discussed below in §2.6.3 .

Following clause stress rule (ii), a speaker can emphasise a particular clause constituent by moving it to the normally stressed pre-predicator position, for example:

( 1 8) Kayangka minya ng'ayku nukany. 'The dog ate illY meat.' cf. (5)

However, any constituent may be stressed for emphasis anywhere in the clause, overruling the unmarked stress pattern of the particular clause type, for example:

(19)	J'awunkarra kaday. ' (Our) friends are coming.'	cf. (1)
(20)	K'irrbaji yalbay. 'The dugong is big.'	cf. (2)

which could be used in answer to the questions 'who is coming?' and 'what is big?'. Further examples of marked stress patterns are:

• (2 1) Mayi m'ayngku nyulu nukany. 'He ate mangoes.'
• (22) Y'ilayku nyulu kadany. 'He came yesterday. '
• (23) Yilayku nyulu k'adany. 'He came yesterday.'
• (24) Ngayu d'iburr kari nukany. 'I did not eat the � (but something else). '
• (25) Wanyangka mayi j'anbal nukany. 'Who ate the guandongs?'
• (26) Yundu j'albu nyajiny? 'Did you see the woman?'
• (27) Nyulu k'irayku dungany? 'Did he go secretly?'

If a polar question contains an adverb, as in (27), stress is in fact usually placed on the adverb, but it may also occur on the verb (see stress rule (iii) b) according to what the speaker enquires about (see Hershberger & Pike 1 970:796).

In tag constructions, which repeat a constituent in nominal form at the end of a clause, after it occurred in pronominal form in its normal clause position, the added noun, which is set off by a slight pause, receives secondary stress. For instance (a comma ',' indicates pause):

• (28) Nyulu ngayku m'inya nukany, kayangka. 'It ate my meat, the dog did.'
• (29) N gayu ny'ungun nyajiny, yinya jalbu. 'I saw her, that woman.'

The same applies, where the subject is given as an 'afterthought', to clarify the reference of a topic noun (thought at first to have a clear referent), for example:

(30) N gayku m'inya nukany, kayangka. 'Ate my meat - the dog. '

In reprise constructions, on the other hand, which prepose a noun phrase that is then followed by a pronoun in normal clause position (see §5.4. 1 ), the preposed constituent receives primary stress, and the stressed constituent of the actual clause receives secondary stress, indicated by ". There is a slight pause after the preposed noun.

• (3 1) K'ayangka, nyulu ngayku m "inya nukany. 'The dog, it ate my meat. '
• (32) Yinya j'albu, ngayu nyungun ny" ajiny. 'That woman, I saw her. '

If the object-NP is preposed, it is usually the verb, and not the pre-predicator constituent, which receives secondary stress.

40 Chapter 2

2.6.3 Intonation contours

Independent statements have intonation contours that peak at clause stress and fall at the end of the clause (see Hershberger & Pike 1 970:802). Thus (3) and ( 1 2), for example, have the intonation contours

Similarly, tag and reprise constructions have the intonation contours:

(36) Kayangka, nyulu ngayku minya nukany. see (31)

When clauses are co-ordinated or subordinated, the intonation pattern is quite different.7 The intonation peak is characterised by higher-than-clause stress pitch, but not loudness. It falls on the final syllable of the clause, followed by a slight pause. The otherwise wordstressed penultimate syllable is usually lengthened, but not stressed. Clause stress in the first clause is normal in pitch and loudness, but weakened in following clauses, for example:

(37) lana wawubajanga ka·dany, mayi wajuny, nu·kany, wunanany. 'They came to the river, cooked food, ate, and slept.'

In subordinated structures, the final rise in the non-final clause carries over into the following clause, usually without a pause in between:

(38) Ngayu wawubajanga dungay kuyu maninka. 'I go to the river to get some fish. '

⁷ See Hershberger and Pike (1970:803) "sequence contour"; their example is unclear in that the final intonation rise and the English gloss suggest a further co-ordinated clause, in which case the intonation contour of the second clause would not be quite as high initially.

Clauses expressing continuous action by several repetitions of the verb have the following intonation contour. Clause stress falls on the first syllable of the first verb, after which pitch remains at the same level until the end with no final drop, and the very last syllable is distinctly lengthened (see Hershberger & Pike 1970:805), for example:

r---------------------------------------------------

(39) lana dungany dungany dungany dunga::ny. 'They went on and on and on.'

(40) Nyulu mayi nukany nukany nukany nuka::ny. 'He ate and ate and ate and ate.'

3

Intonation contours for polar questions and imperatives are similar to those of statements with final falling intonation, but they differ in level of pitch and particularly in the distance between highest and lowest pitch. If we assume three levels of pitch, the three clauses below can be distinguished as follows:

(41) Yundu kaday.

'You are coming.' (statement)

2 1

2 (42) Yundu kaday? 'Are you coming?' (polar question) 1 --

The difference between pitch levels and, particularly, the difference in the final drop in pitch is clearly distinct for these three types of clauses.

3.1 Parts of speech

One can distinguish the following word classes with mutually exclusive membership in Kuku Yalanji:

Nominals - nouns - adjectives - quantifiers Location and time words Personal pronouns Interrogative pronouns Demonstratives Verbs

Particles Interjections.

The criteria for this classification are partly grammatical, partly formal, and partly semantic. Grammatical criteria are based on syntactic function, and inflectional and derivational possibilities. A purely formal aspect is open versus closed word class.

3. 1. 1 Formal and grammatical criteria

All parts of speech except for verbs, particles and interjections could be considered to be nomina Is in the broadest sense because they all inflect for case. However, the feature of open versus closed word class and differences in inflectional systems and forms invite further subcategorisation.

Demonstratives and pronominal parts of speech are closed word classes with symmetrical paradigms that are ordered according to their semantic content like 'here' and 'there' (demonstratives), person and number (personal pronouns), and 'who', 'what' and 'where' (interrogative pronouns). Particles and interjections also constitute closed word classes to which new members cannot easily be added by speakers of the language.

Nominals, location and time words, and verbs are open word classes; that is, there is room for addition by compounding, derivations and borrowing from other languages.

Members of the nominal, pronominal and demonstrative word classes inflect according to a system of syntactic and local cases. Personal pronouns are set off from the other parts of speech in this group in that they have the same form for transitive and intransitive subject versus a different form for transitive object. The other three classes show a formal distinction between transitive subject on the one hand, and intransitive subject and transitive object on the other hand.

Location and time words occur only in local cases which differ to some extent in form from the local cases of the previously mentioned word classes.

Nominals, and location and time words may form transitive and intransitive verb stems by means of derivations.

Verbs are inflected for tense, aspect and mood; other verbal inflections mark clause subordination which is not indicated by any other means. A verb may be modified by an adverbial particle or an adjective in derived adverbial function.

Particles take no inflections of any kind. According to their function they can broadly be divided into two sub-categories: adjectival particles which may act as predicator, and particles with a modifying, often discourse related, function at clause level.

Interjections stand outside the syntactic framework of a clause or sentence. They are usually uttered just by themselves, or preceding a clause or sentence in a discourse context.

3. 1.2 Semantic content

Generally speaking, nominals refer to objects and properties of objects.

Nouns refer predominantly to corporeal objects, such as human beings (including kinship terms, external and internal body parts and bodily excretions), fauna and flora, geographical and geological items, celestial bodies, weather phenomena, artifacts and other objects of human habitat. Further there are terms for 'imaginable' things, such as ghosts and spirits, several specific terms for different types of noises, and terms for human actions such as language, songs, dances and magic practices. Some bodily and mental states are expressed by nouns (warngku 'sleep', kaka 'pain', kuli 'anger', kuru '(female) love or lust'), while others such as hunger, happiness, desire, etc. are expressed by adjectives, verbs or verbal particles.

Kuku Yalanji nouns include a set of generic terms which can occur on their own or may be used as a kind of classifier in conjunction with another noun (see §4.1.1(c)). These are:

bama	'Aboriginal person'
dingkar	'male'
jalbu	'female'
jawun	'unspecified family member' (now generally 'friend')
minya	'edible animal (including birds and fish)'
dikal	'bird'
jarba	'snake'
kuyu	'fish'
mayi	'edible plants (also meal or food in general)'
juku	'tree (classifies anything wooden such as twigs,sticks, building timber, etc.)'
bubu	'country, ground, earth, space'
kulji	'stone, rock, rock formation (but not mountain)'
bana	'water, liquid'

muyar	'air, wind'
baya	'fire (also 'firewood' and classifies anything to do with fire such as smoke, matches, hot ashes)'1
kalka	'spear'
bayan	'camp, house, shelter'
junjuy	'something, anything'

Adjectives denote properties of any nominal referent, for instance size, dimension, value, age, physical and mental characteristics or states, texture, colour, taste, etc. There are three colour terms: ngumbu 'dark', bingaji 'light' and ngalangala 'red, reddish'.

Bingaji also means 'old man', presumably with reference to white hair: the form looks like a comitative derivation with suffix -ji, but it cannot be determined whether the possible root binga means 'light colour' or specifically 'white hair'. N galangala on the other hand could be a reduplicated noun, cf. mulamula 'bright red' derived from mula 'blood', but apparently there is no analysable root ngala.

Note that a number of emotional and physical states are described by noun-verb compounds, e.g. jiba-badiy 'feel sorry, sad (lit. liver-cry)' or wawulay 'feel exhausted' (wawu-wulay 'breath-die').

Quantifiers include numbers nyubun 'one', djambul 'two' and kulurr 'three' and some general terms for quantities such as 'some', 'many', etc. (see §3.2.5).

Location and time words express points of reference in locational and temporal systems, which will be discussed in detail in §3.3.1 and §3.4 respectively.

Semantic content of personal and interrogative pronouns and demonstratives was briefly mentioned in §3.1 .1 and will be detailed further in §3.5, §3.6 and §3.7 respectively.

Verb roots in Kuku Yalanji generally have concrete meaning, referring to motion and rest, perception, voluntary and involuntary utterances, bodily functions, and actions performed on and affecting or effecting an object. Less tangible actions like mental activities are most often expressed by compound verbs. (For more detail see §3.8.2.l, §3.8.5.l and §3.8.5.2).

Particles cover a wide range of meaning, including manner of action, desire, ability, disapproval, negation, uncertainty, conditionality and possibility. The various functions of adverbial, adjectival and discourse particles are discussed in §3.8.6.1 and §3.9.

Interjections comprise yuwu 'yes' and kari 'no', as well as general encouragement, agreement or disagreement. For a full list see §4.1 0.

3.2 Nominal morphology

Nouns, adjectives and quantifiers can be distinguished by their semantic content and also on the grounds of some morphological properties. Comparative and intensity markers jarra 'rather, more' and -baja(ku) 'very' only occur with adjectives and quantifiers, while derivational affixes except possessive, comitative and privative are used only with nouns.

H. Hershberger and R. Hershberger ( 1 982) regard baya as a loan word from English 'fire'. This could be possible according to regular phonological adaptation rules (see §2.3), but has not been otherwise substantiated.

Furthermore, quantifiers are distinct from the other two nominal classes in that they take only case inflections from the 'neutral' set.

3.2. 1 Case inflections - system

Nominal morphology in Kuku Yalanji distinguishes formally between eight cases, some of which fulfil more than one semantic function. The system of cases and their functions is as follows (for a more detailed discussion see §4. 1 .4):

Core syntactic cases

ABSolutive marks the intransitive subject and transitive object of a clause.

ERGative marks the transitive subject.

Peripheral syntactiC cases

INSTrumental marks the implement that is used by an agent in performing an action.

DATive indicates the goal of a purposeful action, the cause for an action or emotion, or a complement of comparison.

Local cases

General Locative marks location 'in', 'on', 'at', etc., the goal of a movement (allative), and goal of three-argument verbs 'show', 'give', etc.

ABLative indicates a general source 'from', which is prior in time or location to the action or state referred to in the clause.

ABESSive indicates a locational source, but one which is not left behind by the actor, i.e. a place from which an action is performed.

PERLative is used only in the Nyungkul dialect and marks a defined space through which someone or something passes. (Yalanji speakers use locative case in this function.)

3.2.2 Case inflections -form

Kuku Yalanji has two sets of inflectional allomorphs for all cases except absolutive, instrumental, abessive and perlative. I have termed these inflectional sets 'neutral' and 'potent'. Allomorphs from these sets may be chosen by a speaker according to whether the referent is perceived to be inherently capable of action ('potent' inflection) or not ('neutral' inflection). This distinction largely coincides with the actual 'animacy' or 'inanimacy' of the referent, but does not necessarily have to. (A full discussion of the semantic properties of neutral and potent case inflections is to be found in §4.1 .4.2.)

Case inflections and their allomorphs are listed in Table 3.1 ; examples of inflected nouns illustrating the different allomorphs are given below. Inflections for which there is no distinction between 'neutral' and 'potent' are listed under 'neutral', since this may be regarded as the unmarked set, whereas 'potent' is the marked set (see §4. 1 .4.2).

'Neutral' case inflections do not receive a special gloss in interlinear morpheme translations of example sentences; 'potent' case inflections are glossed as :pt, as in DAT:pt.

Examples of case inflections and their allomorphs:

46 Chapter 3

Absolutive case inflection is always zero; that is, the citation form of a nominal coincides with its form in intransitive subject and transitive object function, for example:

kaya-ø	'dog'
dingkar-ø	'man'
jarramali-ø	'thunderstorm'

Neutral ergative and instrumental have the allomorphs:

-bu/V-

jna-bu	'foot'
kulji-bu	'stone'
juku-bu	'tree'

• Vbu/r-

wungar-abu	'sun'
badur-ubu	'fishing line'

-njV/y-

balbay-nja	'light, lightning'
diliy-nja	'corkwood pine'
junjuy-nju	'something'

-dV / elsewhere

jalun-du	'sea'
jalbany-da	'taboo food'
bayil-da	'freshwater perch'
balarr-da	'scabies'

Potent ergative has the allomorphs:

-ngkV/V-

kaya-ngka	'dog'
kami-ngka	'father's father, mother's mother'
ngurrku-ngku	'mopoke'

-VngkV/C-

ngangkin-angka	'porcupine'
walkarr-angka	'black goanna'
dingkar-angka	'man'
dubuy-ungku	'small brown kingfisher (messenger bird)'
wabul-ungku	'Torres Strait Pigeon'

	Neutral	Potent
Absolutive	- $\phi$
Ergative	-bu/V- -Vbu/r- -njV/y- -dV/elsewhere	-ngkV/V- -VngkV/C-
Instrumental	= neutral ergative
Dative	-ngkV/y- -VkV/r- -kV/elsewhere	-nkV/V- -VnkV/C-
General Locative	-ngV/V- -V/r- -mbV/y- -bV/elsewhere	-ndV/V- -VndV/C-
Ablative	-muny	-general LOC + muny
Abessive	-mundu
Perlative (Nyungkul dialect)	-burr/V- -Vburr/r- -jVrr/y- -dVrr/elsewhere

Table 3.1: Kuku Yalanji case inflections

Vowel V in the inflection is determined by vowel-harmony rules (see §2.5.1) and is: a after a or i, and u after u.

Neutral dative has the allomorphs:

-ngkVly-

ngujay-ngka	'freshwater turtle'
yibuy-ngku	'loya cane'

-VkVlr-

kadar-aka	‘plains wallaby’
kukur-uku	‘rat’

-kVlelsewhere

ngalkun-ku	'mullet'
dukul-ku	'head'
balji-ka	'dillybag'

48 Chapter 3

Potent dative has the allomorphs:

-nkV/V-

jalbu-nku	'woman'
ngaji-nka	'mother's father'
bama-nka	'Aborigine'

-VnkV/C-

nganjan-anka	father
ngaingai-anka	sand crab
vuru y-mou	spring (water)
wahurr-unku	'shark'

Neutral general locative has the allomorphs:

-ngV/V-

ngara-nga	'roots'
ngarri-nga	'leg'
kiju-ngu	'mud crab'

-V/r-

bibar-a	'shin'
burrir-a	'island'
bujur-u	'feather'

-mbV/y-

buray-mba	'spring (water)'
dajaliy-mba	'deep water'
wakuy-mbu	'upper arm'

-b V / elsewhere

nyabil-ba	'tongue'
dalkan-ba	'casuarina pine'
diburr-bu	'egg'

Potent general locative has the allomorphs:

-ndV/V-

ngiwa-nda	'salt-water eel'
bulki-nda	'cattle'
dunyu-ndu	'husband'
-VndV/C
mukay-anda	'mother's older sister, younger sister's child'
mukirr-anda	'freshwater oyster'
wuyngkul-undu	'spirit of dying person'
kukur-undu	'rat'

Neutral ablative has the form

-muny in all environments, for example:

wawubaja-muny	'river'
manjal-muny	'mountain'
bururr-muny	'crowd, flock'
jalkay-muny	'salmon'
bayan-muny	'house, camp'

Potent ablative has the form

general Locative + muny, for example: ngawa-ndamuny 'baby'

ngamu-ndumuny	'mother'
maral-andamuny	'girl'
diwan-andamuny	'scrub turkey'

Abessive has the form

-mundu in all environments, for example:
'hip'
'grass'
'north'
'sand'

Perlative has the allomorphs:

-burrIV-

walngka-burr	'swamp'
dimbi-burr	'side (of house, etc.)'

-Vburrlr-

burrir-aburr
muyar-aburr

'island' 'wind, air'

-jVrrly-

kabay-jarr duliy-jarr

'small antbed' 'burn grass'

-dV rrl elsewhere

baral-darr wabarr-darr bural-darr buljun-durr

'road' 'shade' 'deserted camping/resting place' 'weeds'

3.2.2. 1 Historical notes

We can recognise a number of reflexes of pA case inflections in Kuku Yalanji. Proto Australian inflections for ergative, purposive (dative) and locative, as reconstructed by Dixon (1 980:3 1 l ff.), are:

ergative:	*-du ~ *-ŋgu ~ *-lu
locative:	*-da ~ *-ŋga ~ *-la
purposive:	*-gu

Reflexes of all but pA *-lu and *-la in Kuku Yalanji are:

neutral ergative	-dV	<	*-du
neutral locative	-ngV	<	*-nga
neutral dative	-kV	<	*-gu
potent ergative	-ngkV	<	*-ngu
potent locative	-ndV	<	*-da
potent dative	-nkV	<	*-gu

As can be seen in the above list, Kuku Yalanji 'potent' locative and 'potent' dative have nasal n before the pA form. It is interesting to note that in some other languages stems of pronouns and human nouns have an incremental n before some non-core case inflections. Warrgamay and Yidiny, for instance, show incremental n with pronouns (Dixon 1980:329 and 1 977:1 68-170 respectively). In Dyirbal proper nouns and some common nouns with human reference may take a stem-forming suffix -J1a- after which the dative inflection is -ngu, whereas it is -gu in other environments (Dixon 1 972:43-44). The fact that non-core inflections in these languages have a special form that is used with human referents, that is referents which definitely possess the 'ability to act', may throw some light on the development of 'potent' inflections for locative and dative in Kuku Yalanji. Specifically 'human' inflections in other languages have been generalised in Kuku Yalanji to apply to all 'potent' referents. But the development of the distinction between a 'potent' and a 'neutral' ergative inflection remains unclear.

3.2.3 Nominal stem formation

Nominal stems in Kuku Yalanji may be formed from nominal roots by affixation, reduplication or compounding.

Stem-forming affixes include the purely formal catalytic suffix -mun-, a nominalising suffix which derives nouns from verbs, and seven suffixes and two prefixes that add a semantic aspect to the meaning of the root. Derivational affixes may be monosyllabic or disyllabic. Criteria for defining these morphemes as derivational affixes are: (a) they cannot occur independently but have to be attached to a nominal root; (b) root and affix are inseparable, that is no other morpheme such as a clitic may occur between them (but two or three derivational affixes may co-occur); ( c) the derived stem may be inflected for case; and (d) the stress pattern defines the derived stem as one word, that is word stress is on the first syllable.

3.2.3. 1 Catalytic Suffix -ffiun-

With trisyllabic nominal roots or sterns and English loan words with any number of syllables (see §3.12) a catalytic suffix which has the form -mun- is inserted between the root and a non-zero inflection. Suffix -mun- does not alter the normal stress pattern of a word as outlined in §2.6.1. Some examples are:

bimakay-mun-ungku	'(not ego's) father's.sister-mun-ERG:pt'
kunjuri-mun-du	'shield-mun-INST'
bulkiji-mun-ku	'pipi-mun-DAT'
mungari-mun-bu	'Kauri.pine-mun-LOC'
kalka-ji-mun-ungku	'spear-COMIT-mun-ERG:pt'

Not all speakers follow this convention consistently, but conservative-minded speakers emphasised that these were the correct forms and leaving out the catalytic suffix would be "not right". (This catalytic suffix is glossed as -mun- in interlinear morpheme glosses.)

3.2.3.2 JDossessive

Alienable possession is expressed by derivational affixes attached to the possessor noun. These suffixes differ according to whether the derived stem has zero inflection (absolutive case) or non-zero inflection; the former also show formal differences in the Yalanji and Nyungkul dialects.

Inalienable possession, such as body parts, may also be indicated by these derivational suffixes, but normally possessor and inalienably possessed noun are simply juxtaposed in a noun phrase without possessive affixation (see §4.5.2). The Yalanji and Nyungkul forms of possessive suffixes are listed in Table 3.2.

	Yalanji	Nyungkul
+ zero inflection	-mu/V-	-ngV/V-
	-Vmu/C-	-VngV/C-
+ non-zero inflection	-ndVmun-/V-	-ndVmun-/V-
	-VndVmun-/C-	-VndVmun-/C-

Table 3.2: Possessive suffixes in Yalanji and Nyungkul

The posseSSIve suffix that is used preceding non-zero inflection could possibly be analysed as

potent locative + catalytic suffix -mun-.2

²For catalytic suffix -mun- see §3.2.3.1. The stem-forming possessive suffix with non-zero inflection in Guugu Yimidhirr is - :ga + -mu-. -mu- is a catalytic suffix like Kuku Yalanji -mun-, and - :ga is interpreted by Haviland ( 1979a:56) as being related to either abessive or goal. (In Djabugay a catalytic suffix, usually -mu-, is added to the regular possessive stem before a non-zero inflection (patz 1991 :269).

Apart from the formal similarity between locative and possessive (see also §3 .2.4), the two constructions also overlap to some extent in function (see §4.5.5.) . .

Some examples of possessive noun phrases in absolutive case in both dialects are:

Yalanji	Nyungkul
kambi ngamu-mu	kambi ngamu-ngu	'mother's clothes'
mayi kaya-mu	mayi kaya-nga	'the dog's food'
dambal maral-amu	dambal maral-anga	'the girl's shoes'
bayan dingkar-amu	bayan dingkar-anga	'the man's shoes'

Non-zero inflections on possessive stems are invariably from the neutral set. Some examples in sentences are:

(44)	Jalbu-ndumun-du	dunyu-ngku	bikibiki	kuni-ny.
	woman-POSS-ERG(A)	husband-ERG:pt(A)	pig.ABS(O)	hit/shoot-PAST
	'The woman's husband shot a pig.'

• (45) N gayu dunga-y bayan-ba I sg.NOM(S) go-NONPAS[ house-LOC 'I am going to granny's house.' kami-ndamun-bu. FFIMM-POSS-LOC
• (46) N gamu-ngku maral-andamun-bu kuyu mani-ny. mother-ERG:pt(A) girl-POSS-ERG(A) fish.ABS(O) get-PAS[ 'The girl's mother caught a fish.'
• (47) Nyulu yinil kaya-nka karrkay-andamun-ku. 3sg.NOM(S) afraid dog-DAT:pt child-POSS-DAT 'He/she is afraid of the child's dog.'

Apparently there is no separate affix to express 'possessor of a possessor' in either dialect. Such constructions, which were only obtainable in elicitation and are therefore probably forced or stilted, show the normal possessive suffix on both possessor nouns, for example:

• (48) Dingkar-amu ngamu-mu bayan yalbay. man-POSS.ABS(S) mother-POSS.ABS(S) house.ABS(S) big 'The man's mother's house is big.'
• (49) Nyulu kada-ny maral-andamun-muny yaba-ndamun-muny 3sg.NOM(S) come-PAS[ girl-POSS-ABL eB-POSS-ABL bayan-muny. house-ABL 'He came from the girl's elder brother's house.'

For a full discussion of possessive constructions, see §4.5.

3.2.3.3 Comitative

Like most Australian languages Kuku Yalanji has a derivational suffix which indicates, broadly termed, 'with noun' or 'having noun'. A comitative form in Kuku Yalanji may have either attributive function or indicate 'accompaniment' of a person in motion or rest. (For a detailed discussion see §4.6.1 .)

Like case inflections, the comitative suffix has a neutral and a potent variant. Furthermore, slightly different forms are used, if the derived stem takes zero inflection or non-zero inflection. The forms of comitative suffix are set out in Table 3.3. These allomorphs have been used most consistently in elicitation and are generally regarded as the correct forms. However, in normal speech speakers tend to use -iji and -ji in free variation after a rhotic, and forms with or without final -rr in free variation before any inflectional ending.

	Neutral	Potent
+ zero inflection	-iji/R-	-nji/V-
+ non-zero inflection with initial n or ng	-ji/elsewhere	-Vnji/C-
+ non-zero inflection	-ijirr/R-	-njirr/V-
with initial stop or m	-jirr/elsewhere	-Vnjirr/C-

Table 3.3: Allomorphs of comitative suffix in Kuku Yalanji
	-da
	-ga
	-bada
	-baga

These forms are obviously cognate to the comitative suffixes in the genetically related languages: Yidiny -ji ,.." -yi, Djabugay -i ,.." -nyji ,.." - :rr, and Guugu Yimidhirr -dhirr. It appears that the Guugu Yimidhirr suffix is the most conservative form (cf. -girri in Warrgamay, further south than Yidiny, -dharri in Wirandharri, a Victorian language (Dixon 1 980:325)), whereas the related languages have generally dropped the final -rr but for a few allomorphs, e.g. - :rr/V- in Djabugay, and the suffixes preceding some non-zero inflections in Kuku Yalanji. But even in this environment Kuku Yalanji seems to be gradually losing the final -rr.

Some examples of comitative stems with zero inflection are:

Neutral suffixes

bulbur-iji	'with dust, dusty'
jukar-iji	'with sand, sandy'
walarr-iji	'with a beard'
juku-ji	'with a stick'
kaka-ji	'with pain, painful'
milbal-ji	'with tears, tearful'
ngunyin-ji	'with a dillybag'
Potent suffixes
yaba-nji	'with elder brother'
yabuju-nji	'with younger brother'
nganjan-anji	'with father'
jawun-unji	'with a friend'
kadar-anji	'with a wallaby'

Some examples of comitative stems with non-zero inflection are:

(50)	Ngayu 1sg.NOM(S)	jalbu-ndu woman-LOC:pt	dunga-y go-NONPAST	ngawa-nji-nda. baby-COMIT-LOC:pt
	'I go to the woman with the baby.'

• (5 1) Warru-ngku kuli-ji-ngka kamukamu nuka-ny. yg.man-ERG:pt(A) rage-COMIT-ERG:pt(A) 'The wild young chap drank alcohol.' alcohol.ABS(O) eat/drink-PAST
• (52) N gayu binal dingkar-anka I sg.NOM(S) know man-DAT:pt 'I know the man with the spear. ' kalka-jirr-ka. spear-COMIT-DAT
• (53) Nyulu warri-ny bubu-muny dubu-njirr-muny. 3sg.NOM(S) run-PAST place-ABL ghost-COMIT-ABL 'He ran from the place with the ghost. '

More examples are given in §4.6. 1.

3.2.3.4 Privative

A nominal stem meaning 'without noun' or 'noun-less' may be formed by suffixing the negative particle kari to a nominal root. The resulting stem is clearly definable as one word because of the stress pattern (only the first syllable is stressed), and because a non-zero inflection can only be attached after -kari-.

Privative stem formation is far less productive than comitative stem formation (see §4.6.2) and privative stems are seldom used in other than equational sentences, so a privative stem rarely occurs with non-zero inflection. Some examples of privative stems in equational sentences are:

• (54) Yinya bama milka-kari. that.ABS(S) Aborigine.ABS(S) ear-PRIV 'That Aborigine is deaf.' (lit. no ears)
• (55) Ngayu mani-kari. I sg.NOM(S) money-PRIV 'I have no money.'
• (56) Juku-juku nganka-kari. tree-REDUP.ABS(S) flower-PRIV 'The trees have no flowers.'

Some examples of privative stems with non-zero inflections are:

• (57) Kaya-kari-ngka bikibiki kari warmba-bunga-ny. dog-PRIV-ERG:pt(A) pig.ABS(O) NEG uncovered-CAUS-PAST 'The one without a dog found no pigs'.
• (58) Ngayu wawu bangkarr-ka bajibay-kari-ka. I sg.NOM(S) want meat/steak-DAT bone-PRIV-DAT 'I want meat without bones.'
• (59) Jana kambi daji-ny maral-anda mani-kari-nda. 3pl.NOM(A) clothes.ABS(O) give-PAST girl-LOc:pt money-PRIV-LOC:pt 'They gave clothes to the girl with no money.'

Some speakers prefer neutral inflections with privative sterns, e.g. kaya-kari-bu in (57) and maral-anda mani-kari-ng.a in (59).

3.2.3.5 Number

A simple root by itself generally implies a singular referent in Kuku Yalanji (except for mass nouns), for example:

jalbu	'a woman'
wawubaja	'a river'

In context, a plural interpretation of a simple root is possible, as for instance in (65) below: nganjirr 'grass trees'. To specify more than one referent a quantifier (see §3.2.5) may be used together with a simple root, for example:

bama wubul	'many Aborigines'
jarba jambul	'two snakes'

Apart from this there are three derivations which can indicate plural or dual.

(i) Reduplication

Reduplication of the noun root is the normal device to imply general plural, for example:

wulman-wulman	'old men'
kangkal-kangkal	'own children'
kumu-kumu	'mosquitoes'
bilngkumu-bilngkumu	'salt-water crocodiles'
juku-juku	'trees'

• (60) Baral-baral duna julbarr. pathlroad-REDUP.ABS(S) wet slippery 'The roads are wet and slippery. '
• (6 1) Maral-maral-angka bulkiji wubul baka-ny. girl-REDUP-ERG:pt(A) pipi.ABS(O) many.ABS(O) dig-PAST 'The girls dug up a lot of pipis.' (a type of small salt-water mussel)

(Adjectives are reduplicated to fulfil adverbial function (see §3.8.6.2.).)

(li) Kinship plural

Kinship terms (including jawun 'unspecified relative, friend' but excluding kangkal 'own child') are not reduplicated for plural, but take the derivational suffix -karra- instead, for example:

dunyu-karra	'husbands'
yabuju-karra	'younger brothers'
ngaji-karra	'maternal grandfathers'
ngalayan-karra	'fathers-in-law'

(62) Nyulu jawun-karra-nda kunja-n-kunjaji-ny mayi-ka. 3sg.NOM(S) friend-KPL-LOC:pt ask.for-n-REDUP-PAST food-DAT 'He was asking around for food among his friends.'

(63) Jana-nda 3pl-LOc:pt mani-nka!" get-pURP manyarr-karra-ngka yalama-ny: wife-KPL-ERG:pt(A) say-PAST 'The wives said to them: "Go to get some fish!'" "Dunga-y kuyu go-IMP fish.ABS(O)

The cognate 'collective plural' suffix -garr in Guugu Yimidhirr has wider application than the Kuku Yalanji suffix in that it can occur with any noun, e.g. guda-garr 'dogs'. On the other hand, with kinship terms it seems to have a more restricted meaning in that it implies "that several people stand in the same relation to a single other" (Haviland 1 979a:55), e.g. gaarga-garr 'younger brothers of a single person'. In Kuku Yalanji -karra apparently does not imply relationship to just one person, but simply indicates plural, as for instance in (63), and:

(64) N gamu-karra nganjan-karra school-mun-bu dunga-y mother -KPL.ABS(S) father -KPL.ABS(S) school-mun-LOC go-NONP AST

teacher-mun-unji balka-wa-nka.

teacher -mun-COMIT:pt talk -RECIP-PURP

'The mothers and fathers go to the school to talk with the teacher.' (i.e. the parents in the community, not just the parents of a single person)

(iii) Derivational suffix -bulal 'pair'

This derivational suffix is obviously related to the dual pronoun bula '3. person dual' (see §3 .5.1). It is most frequently used with nouns with human reference and the noun phrase often also includes the dual pronoun and/or the quantifier jambul 'two' as for instance in text 51, line 1 and:

(65) Bula jalbu-bulal-angka dubu waju-ny nganjirr-da. 3du.NOM(A) woman-pair-ERG:pt(A) ghost.ABS(O) bum-PAST grass tree-INST 'The two women burnt the ghost with [burning] grass trees.' (from a story about two women avenging the theft of their children by a ghost)

The use of -bulal implies some intrinsic link between the two referents, that they somehow belong together and act together. To indicate simply 'two', without wanting to imply some connection between the referents, the quantifier jambul 'two' is used.3

Consider:

• (66) a. Yarraman-bulal kankada-y. horse-pair .ABS(S) come.REDUP-NONP AST 'The pair of horses is coming.' (e.g. a mare and her foal that are always together)
  • b. Yarraman jambul wawubaja-nga wanarri-y. horse.ABS(S) two.ABS(S) river-LOC run.REDUP-NONPAST 'Two horses are running to the river.' (they don't belong together)

3 Compare the use of derivational suffix -dpran 'two, each of two' and adjective bulay 'two' in Dyirbal (Dixon 1 972:228-229)

• (67) a. luku-bulal ymyay janjana-y. tree-pair.ABS(S) there stand.REDUP-NONPAST 'A pair of trees is standing there. ' (e.g. standing close together and perhaps of the same kind and size)
  • b. Ngayu jambul juku nyaji-ny. I sg.NOM(A) two.ABS(O) tree.ABS(O) see-PAST 'I saw two trees.' (e.g. one by the roadside and one on the beach)

3.2.3.6 Nominalising SUffix -nyV

A noun stem may be derived from a verb by affixation of -nya after final vowels a or i, or -nyu after final vowel u.4 This suffix is identical in form with the unmarked verbal inflection with subordinating discourse function in the Nyungkul dialect and the 'causal' subordinate inflection (see §3.8.4.2 (viii)) and a clear distinction between -nyV as a verb inflection and -nyV as a nominaliser is not possible. For instance, a -nyV form may have habitual implication (see H. Hershberger & R. Hershberger 1 982:284) which could well represent a nominal form:

(68) Nyulu minya nuka-nya. 'He always eats meat. ' OR 'He is a habitual meat-eater.'

But I have not encountered constructions like this with further case inflections, as for instance 'the habitual meat-eater-ERG shot a pig'. In my experience, -nyV stems with further case inflections occur only in 'simultaneous', 'prior event' and 'hypothetical reason' subordinate clauses (see §4.4.4). In these uses nominal stems with -nyV have basically a subordinating role, i.e. they are functionally similar to verb + -nyV forms in 'causal' subordinate clause (see §4.4.4.2) and in subordinate 'non-event' marking in discourse (see §5.5.1 .2). Neither of these can take any case inflections. It may therefore be possible that verbal inflection -nyV and nominal stem-forming suffix -nyV are related.

F. Merlan (1 98 1) addresses a similar problem in Mangarayi, a language of the western Roper River area, where "The same verbal prefix forms are used to mark: (i) irrealis mood; (ii) generalised subordinate clauses ( ... ); and (iii) an inflectional 'habitual' category of the verb ( ... )" (1981:1 75). She concludes that, "Basically, the shared function of such markers is to signal that the basis for interpretation of the predicates to which they are affixed is not given there, but is to be found within the fairly immediate linguistic context" (1981 :208). Thus the function of this Mangarayi prefix seems to be similar to the general subordinating function of suffix -ny V in Kuku Yalanji.

3.2.3. 7 Other noun derivations

There are some derivational affixes in Kuku Yalanji that add, broadly termed, a semantic aspect of 'personal characteristic' to a noun root or stem. These are -( w)arra 'inhabitants of', -baka 'excessive' and mala- 'expert'.

⁴ For vowel-harmony rules see §2.S.1

(i) -(w)arra 'inhabitants of'

This suffix attaches to proper place names or comitative stems referring to characteristics of larger areas (see §1.2. 1 (i) for area distinctions in Kuku Yalanji) and has the meaning ‘inhabitant(s) of X or native of X’. Its allomorphs are:

-warra/V -arra/C-

In Kuku Yalanji this derivational suffix also apparently implies plurality, i.e. there is no attested use of -(w)arra in reference to just one person, for example:

jalun-ji-warra	'seaside people'
mangkal-arra	'Cedar Bay people'

• (69) K ubirri-warra wujalwujal-arra-nda (Mossman)-inhab.ABS(S) (Bloomfield)-inhab-LOC:pt 'The Mossman people went to the Bloomfield people. ' dunga-ny. gO-PAST
• (70) Maja-ji-warra-ngka kurranji scrub-COMIT -inhab-ERG:pt(A) cassowary.ABS(O) 'The scrub people killed many cassowaries. ' wubul kuni-ny. many.ABS(O) hitJkill-PAST

This suffix has cognates in all the genetically related languages (warra in Guugu Yimidhirr, -barra in Djabugay and Yidiny) as well as in many other languages of southern and central Queensland (see Tindale 1 974).

Haviland ( 1 979a:64) has interpreted warra in Guugu Yimidhirr as an "independent particle with nominal expression", but in Kuku Yalanji -(w)arra is clearly a suffix, as is shown by the allomorphic variation according to the final root or stem segment.

(ii) -baka 'excessive (characteristic)

Suffix -baka attached to a simple or reduplicated noun root, usually with abstract reference, indicates that the reference of the noun applies excessively to someone, i.e. 'too much noun' or 'excessively noun', which has always negative connotations. The derived stem generally functions as a noun rather than an adjective and is often used as a kind of proper noun in reference to a person, for example:

kuli-baka rage-excess 'excessively wild/grumpy person or animal'

jalbu-baka jalbu-jalbu-baka woman-REDUP-excess 'notorious womaniser'

kuru-baka fem.lovellust -excess 'flirtatious, "easy" woman'

ngirray-baka weeping(noun)-excess 'crybaby'

As can be seen in the above examples, a noun root in connection with -baka may have an extended meaning. For instance, jalbu-baka is not an excessively feminine person but a lecherous male. The semantic extension can go even further, for example:

dirra-baka tooth-excess 'someone who swears too much'

Some examples of these derivations in sentences are:

• (7 1) N gayu yinil kuli-baka-nka. ¹sg.NOM(S) afraid rage-ex cess-DA T:pt 'I am scared of the wild one.'
• (72) Nyulu jalbu-baka-ngka maral dumbarri-ny 3sg.NOM(A) woman-excess-ERG:pt(A) girLABS(O) break-PAST buyun-dama-ny. bad-CADS-PAST 'The womaniser raped a girL '

An 'excessive' nominal stem may also be used with no reference to anyone in particular, but as a general lament about a deplorable situation, for example:

(73) Kamu-kamu-baka! 'There is too much drinking going on around here!'

(74) Nyurra-baka! 'There is too much noise around here!' (nyurra 'irritating noise of people shouting and laughing’)

(iii) Mala- 'expert'

A nominal stem formed with mala- attached to a noun root has the semantic content ‘good at noun’ or ‘noun expert’, for example:

mala-kalka	'spear expert'
mala-minya	'good hunter, game expert'
mala-mayi	'good cook, food expert'
mala-burri	'good singer, song expert' ( burri ‘name, corroboree song’)

• (75) Charlie mala-bulki. Charlie.ABS(S) expert-cattle 'Charlie is an expert stockman.'
• (76) Nyulu mala-mayi-ngka 3sg.NOM(A) expert-food-ERG:ptCA) 'The good cook cut up the fish.' kuyu yaka-ny. fish.ABS(O) cut-PAST
• (77) Mala-balji-nda babaji-ka balji-ka. expert-dillybag-LOC:pt ask.about-IMP dillybag-DAT 'Ask the dillybag expert about diUybags !'

The original meaning of this prefix is probably 'hand' since there are cognates mala 'palm of hand' in Yidiny and 'hand' in Dyirbal, south of Yidiny (cf. English 'he is handy atlhas a hand for woodwork'). The word for 'hand' in Kuku Yalanji and Djabugay is mara, obviously cognate to the Yidiny and Dyirbal forms. While mara is a free noun root, mala- cannot occur on its own in Kuku Yalanji. It is interesting to note that the intonation pattern of mala- stems is not consistent. Sometimes the first stem syllable is stressed, e.g. mala-minya, which concurs with the normal word-stress pattern, but more often only the first root syllable is stressed, e.g. mala-m{nya. This might indicate that mala- is in fact treated as a pro-clitic rather than a stem-forming prefix.

One Nyungkul speaker also used mala- with nominalised verb stems (see §3.2.3.6), e.g. mala-dunga-nya 'good at getting away' (text 1 2, line 1 8) or mala-bundanda-nya '(always) sitting comfortably (and unworried), (text 1 2, line 55). This use of mala- was not confirmed by other speakers, who only attach it to noun roots.

3.2.3.8 'Degree' of adjective

There are two affixes which occur only with adjectives or comitative stems in attributive or adjectival function (for comitative see §4.6. 1 ). They indicate the degree or intensity of the property described by the adjective.

(i) Jarra- 'rather, more'

Jarra- preceding an adjective indicates 'fairly' or 'rather adjective', for example:

jarra-yalbay	'rather big'
jarra-minday	'rather tame'
jarra-jinbal	'rather fast'
jarra-bambay	'rather sick'

(78)	Nyungu	kangkal	jarra-kaykay.
	3sg.POSS.ABS(S)	child.ABS(S)	DEG1-small
	'Her child is rather small.'

• (79) Nyulu bunda-ny mayngku-ngu jarra-kulbul-bu. 3sg.NOM(S) sit-PAST mango-LOC DEGl-ripe-LOC 'He sat on a rather ripe mango. '
• (80) Kaya-ngka jarra-kuli-ji-ngka ngayku dog-ERG:pt(A) DEGl -rage-COMIT-ERG:pt(A) lsg.POSS.ABS(O)

kaya bayka-ny. dog.ABS(O) bite-PAST 'A rather wild dog bit my dog.'

Stress in jarra- stems is assigned according to regular word-stress pattern on the first syllable, e.g. jarra-yalbay.

Adjectival stems with jarra- are also used in comparisons in which case jarra- indicates 'more adjective than'. The compared component takes dative case (see §4.l AA (ii)), for example:

(8 1) Minya ngawuya jarra-kima minya-ka bulki-ka. meaLABS(S) sea. turtle. ABS(S) DEGl-soft meat-DAT cattle-DAT 'Sea-turtle meat is softer than beef. '

J arra- has also been attested with one verb, jarra-kulba-l 'carry' (kulba-l 'lift'), but it is not used productively with verbs.

(ii) -Baja (-ku) 'very'

The bound morpheme -baja has various functions. With verbs it is used as a postinflectional clitic meaning 'again' (see §3 . 1 0.3). With adjectives, on the other hand, it may be more appropriately classed as a stem-forming suffix for the following reasons. In connection with an adjective, -baja must always be followed by the emphatic clitic -ku (see §3. 1 0.2) to indicate 'very adjective', for example:

• (82) N gayu dakuy-baja-ku. I sg.NOM(S) hungry-DEG2-EMPHl 'I am very hungry.'
• (83) Ngawa kingkin-baja-ku. baby.ABS(S) cute-DE�-EMPHI 'The baby is very cute.'

If, however, an adjectival stem with -baja takes non-zero inflection, the inflectional suffix is inserted between -baja- and clitic -ku, for example:

(84)	Dingkar-angka junkurr-ji-baja-ngka-ku man-ERG:pt(A) strength-COMIT-DEG2-ERG:pt(A)-EMPH1
	bikibiki kulba-ny. pig.ABS(O) lift-PAST 'The very strong man lifted the pig up.'
(85)	Naavu iha-hadi-v karrkay-anka hambay-ho

(85) Ngayu jiba-badi-y karrkay-anka I sg.NOM(S) liver-cry-NONPAST child-DAT:pt 'I am sorry for the very sick child.' bambay-baja-nka-ku. sick-DEG2-DAT:pt-EMPH2

3.2.3.9 Co-occurrence of stem-forming affixes

The most productive of all stem-forming processes are possessive, cornitative, and general plural (reduplication). Adjectival stem formation with 'degree'-affixes is also fairly common, but all other stem-forming affixes have more limited application, partly because of their semantic content and partly because of formal criteria. For instance, 'kinship plural' -karra is used only with kinship terms and 'inhabitants' suffix -(w)arra can only occur on proper or common place names. Similar reasons proscribe the co-occurrence of derivations like 'plural' and 'pair'.

The catalytic suffix -mun- is used only with trisyllabic roots and loan words preceding the case inflection and therefore does not have to be included in ordering rules of stem-forming suffixes. The two prefixes also do not have to be ordered, since mala- 'expert' occurs only with noun roots andjarra- 'rather, more' attaches only to adjectival stems.

Of the stem-forming suffixes which occur with nouns only, 'privative', 'kinship plural', 'pair', 'inhabitants', and 'excessive' can be classed as first-order suffixes; reduplication for 'general plural' also falls into this category. All of these may be followed by comitative in 'accompaniment' function or possessive, or both, for example:

• (86) N gayu jilba dunga-y kangkal-kari-ji. Isg.NOM(S) walk go-NONPAST own.child-PRIV-COMIT 'I go for a walk with the childless one.'
• (87) lana kada-ny-baja ngamu-karra-ndamun-ji kambi-ji. 3pl.NOM(S) come-PAST-again mother-KPL-POSS-COMIT clothes-COMIT 'They came back with the mothers' clothes.'
• (88) lalbu-bulal-amu kangkal dubu-ngku ngaki-ny. woman-pair-poss.ABS(O) own.child.ABS(O) ghost-ERG:pt(A) steal-PAST 'A ghost stole the two women's children.'
• (89) Banabila-warra-mu kaya kuli-ji. (place name)-inhab-POSS.ABS(S) dog.ABS(S) rage-COMIT 'The Banabila-people's dogs are fierce.'
• (90) Kari dunga-y yinya-nji kamu-baka-nji. NEG go-IMP that-COMIT:pt a1cohol-excess-COMIT:pt 'Don't go with that drunkard!'
• (9 1) Maja balka-n-balka-wa-y warru-warru-nji. boss(master).ABS(S) talk-n-REDUP-RECIP-NONPAST yg.man-REDUP-COMIT:pt 'The boss is talking with the young men.'

A comitative stem in attributive function may also be followed by possessive:

(92) lalbu-mu waral-ji-mu dunyu wula-ny. woman-POSS.ABS(S) belly-COMIT-POSS.ABS(S) husband.ABS(S) die-PAST 'The pregnant woman's husband died.'

As was shown in §3.2.3.7 (i), an 'attributive' comitative stem referring to geographical features may take the 'inhabitants' suffix, for example:

ngalkal-ji-warra

'open forest country people'.

A comitative stem in 'accompaniment' function, on the other hand, cannot take further derivational suffixes, but such a comitative stem may be based on a possessive stem, as for instance in (87), and

(93) Nyulu warri-ny buliman-andamun-ji marrkin-ji. 3sg.NOM(S) run-PAST police-POSS-COMIT rifle-COMIT 'He ran [off] with the policeman's rifle.'

Note that possessive and privative apparently cannot co-occur. A sentence like 'he ran off without the policeman's rifle' would be paraphrased as 'he did not run off with the policeman's rifle', using a comitative form and negating the clause.

The adjectival 'degree' suffix -baja(-ku) may also be followed by further derivations, but such constructions are unusual, for example: .

• (94) Dingkar-amu yalbay-baja-mu-ku kalka kadaba-ny. man-POSS.ABS(S) big-DEG2-POSS.ABS(S)-EMPHI spear.ABS(S) break-PAST 'The very big man's spear broke.'
• (95) Nyulu dunga-ny car-mun-ji mukul-baja-ji-ku. 3sg.NOM(S) go-PAST car-mun-COMIT old-DEG2-COMIT-EMPHI 'He went with a very old car. '

Co-occurrence possibilities of nominal derivational suffixes are listed in Table 3 .4.

		may be followed by
Privative	-kari
Kinship plural	-karra
Pair	-bulal
Inhabitants	-(w)arra	Possessive -mu ~ -ngV
		AND/OR
Excessive	-baka	'Accompaniment' comitative -ji ~ -nji*
General plural	reduplication
Degree₂	-baja(-ku)
'Attributive' comitative	-ji ~ -nji	Possessive -mu ~ -ngV OR Inhabitants -(w)arra
Possessive	-mu ~ -ngV	'Accompaniment' comitative -ji ~ -nji
		* For further allomorphs of possessive and comitative see §3.2.3.2 and §3.2.3.3 respectively.

	-bana	-danyu	-diji	-dina	-gambay	-jarranga	-ji	-jima	-jin	-kala	-ngun	-rr	-rrba	-tharra	-waji
-bana
-danyu
-diji
-dina
-gambay
-jarranga
-ji
-jima
-jin
-kala
-ngun
-rr
-rrba
-tharra
-waji

3.2.3. 10 Nominal compounds

Nouns and adjectives can be formed by compounding roots or stems, which often results in idiomatic expressions with extended semantic content. A compound is defined by inseparability and word-stress pattern (see §2.6.1 ).

Some compounds cannot be analysed synchronically, for example:

marrka-bina	'oyster found in mangrove swamp' ( marrka 'salt pan' in Kuku Yalanji, bina 'ear' in Djabugay, which may be a reference to the shape of the shell)
kalka-muku	'green tree snake' (lit. spear-back; this compound probably refers to the long and slender shape of the reptile)

More productive compounding processes include body-part terms, particularly miyil 'eye' and milka 'ear', for example:

miyil-burra	'blind' (eye-stale/bitter)
miyil-dudu	'illiterate' (eye-blunt)

miyil-kaban-ji	'literate' (eye-letter/paper-COMIT)
miyil-wurril	'cross-eyed' (eye-lopsided)
miyil-wujurr-wujurr	'dusk' (eye-darkness-REDUP)
milka-dudu	'deaf' (ear-blunt)
milka-bujar	'homesick' (ear-raw/cold)
milka-marri	'contented, well adjusted' (ear-grown.up)
milka-dandi	'good memory' (ear-hard)
walu-dandi	'stubborn, disobedient' (face-hard)
mara-jalajala	'generous person' or 'clumsy person' (hand-loose/broken)
manu-buray	'squeamish feeling' (throat-spring.water)

The use of body-part terms in nominal and verbal compounds (see §3.8.S.2) with reference to personal attributes or mental states has also been attested for Guugu Yimidhirr (Haviland unpublished word list) and Djabugay (Patz, unpublished word list).

3.2.4 Morphology of kinship terms

As was shown in §3.2.3.5 (ii) kinship terms differ from other nouns in that they take a special derivational suffix to indicate plural. A subset of seven kinship terms out of an attested total of 31 is further distinguished from other nouns in that they take case inflections and comitative suffixes from the 'neutral' set. These are:

kangkal	'own child'
manyarr	'wife'
babarr	'older sister'
jinkurr	'younger sister'
biwul	'mother-in-law'
ngalayan	'father-in-law, son-in-law'
kujiway	'daughter-in-law'

Of these, kangkal 'own child' always occurs with 'neutral' suffixes. With the other kinship terms 'neutral' suffixes are preferred, but 'potent' suffixes are possible, for example:

(96)	Kangkal-da(*-angka)	kaya	kuni-ny.
	own.child-ERG(A)	dog.ABS(O)	hit-PAST
	'[My] child hit the dog.'

• (97) Manyarr-dal-angka mayi waju-l. wife-ERGI-ERG:pt(A) food.ABS(O) cook-NONPAST 'The wife is cooking food.'
• (98) Ngayu dunga-y jinkurr-bul-undu. I sg.NOM(S) go-NONPAST yZ-LOO-LOC:pt 'I go to [my] younger sister. '
• (99) Nyulu kada-ny biwul-jil-unji. 3sg.NOM(S) come-PAST MiL-COMIT/-COMIT:pt 'She came with [her] mother-in-law. '

Other kinship terms allow only 'potent' suffixes, for example:

• ( 1 00) Dunyu-ngku(*-bu) ngawa kuji-l-kuji-l. husband-ERG:pt(A) baby.ABS(O) look.after-REDUP-NONPAST 'The husband is looking after the baby. '
• (101) Ngayu yaba-nda(*-nga) babaji-ny. I sg.NOM(S) eB-Loc:pt ask-PAST 'I asked [my] elder brother.'

In the Yalanji dialect the same kinship terms also prefer a possesslve suffix which is identical to the locative. inflection, for example:

kambi babarr-bal -amu 'elder sister's clothes' kambi kujiway-mbal-amu 'daughter-in-Iaw's clothes'

In the Nyungkul dialect, where the possessive suffix already coincides with allomorphs of the neutral locative inflection, there is no distinction between possessive stems of these kinship terms and other nouns, for example:

kaya manyarr-anga	'the wife's dog'
kaya biwul-ungu	'mother-in-law's dog'
kaya maral-anga	'the girl's dog'

3.2.5 Morphology of quantifiers

Kuku Yalanji has the basic two-term number system:

nyubun	'one'
jambul	'two'

There is a synonym mamarra 'two' which, according to my informants, is an "oldfashioned" or "stylish" form, but not a taboo language item. Although widely known, this term is not in general use.

Further quantifiers are:

kulur	'a few, three'
kanbal	'some, several'
wubul	'many'
kankur	'half'
yambayamba	'everything'
yamba-kari	'nothing'

Kulur 'a few' now generally refers to cardinal number 'three' (see H. Hershberger & R. Hershberger 1 982), but it may also be used to imply an unspecified small number larger than two and smaller than kanbal 'some'. Two more derived number terms are:

jambul-jambul	'four (two-two)'
mara	'five (hand)'.

There are also two quantifiers which refer both to size and, by extension, to uncountable amounts. These are:

buban	'(a) little, insufficient'
jirray	'big, plenty, sufficient'

The following sentences illustrate the use of these two tenus in contrast to their counterparts kanbal and wubul:

• (102) a. Kambi ngayku kanbal. 'I have several [items of] clothes.'
  • b. Kambi ngayku buban. 'I have too few clothes/my clothes are insufficient.'
• ( 1 03) a. Ngayku bulawu kanbal. 'I have some flour. '
  • b. Ngayku bulawu buban. 'I have only a little flour. '
• ( 1 04) a. Bayan ngayku wubul. 'I have many houses.'
  • b. Bayan ngayku jirray. 'My house is big enough.'
• (105) a. Birra wubul juku-ngu. 'There are many [identifiable] leaves on the tree[s]
  • b. Birra jirray juku-ngu. 'There is a mass of leaves on the tree[s]' or: 'There are large leaves on the tree [ s]'

but c. Birra yalbay (bigllarge)juku-ngu.

can only mean 'There are large leaves on the tree(s)'.

As can be seen from the above examples the use of 'mass' quantifier versus 'countable' quantifier does not depend on the type of noun, but the semantic distinction is inherent in the quantifier. Consider also:

bama wubul	'many people'
bama jirray	'a large crowd of people'
juku wubul	'many trees'
juku jirray	'a large forest'

Quantifiers have the same system of case inflections and derivational possibilities as other nominals. But they differ formally from nouns and adjectives in that they apparently take only 'neutral' suffixes. Furthermore, cardinal numbers one and two in the Nyungkul dialect, and one in the Yalanji dialect take case inflections and comitative and possessive suffixes that are different from those of other nominals. These suffixes are listed in Table 3.5.

Kulur 'a few, three' in the Nyungkul dialect takes these suffixes for ergative and ablative and, in free variation with regular nominal suffixes, also for dative and possessive. Comitative has the regular 'neutral' form -iji, i.e. kulur-iji (see Table 3 .3). Some speakers also use the cardinal number suffixes in free variation with regular suffixes for quantifier wubul 'many'.

Ergative, dative and locative inflections without initial vowel i are identical to the neutral allomorphs for these cases after y-final roots (see Table 3.1). It would therefore be possible that vowel i is in fact not part of the suffix, but part of a suppletive root ending in -iCy). Such an assumption would be supported by the fact that vowel i remains constant, whereas the suffix-final vowel in these inflections is a after root vowel i, according to the vowel-harmony rule. However, there is no other synchronic or diachronic evidence for suppletive trisyllabic roots of cardinal numbers. For this reason I interpret vowel i as part of the suffix and not as part of the root.

		'one'	'two'
Absolutive	-∅	nyubun	jambul
Ergative	-inja	nyubuninja	jambulinja
Dative	-ingka	nyubuningka	jambulingka
Locative	-imba	nyubunimba	jambulimba
Ablative	-imuny	nyubunimuny	jambulimuny
Comitative	-inji	nyubuninji	jambulinji
Possessive	=locative
		(Abessive and perlative have not been attested)

Table 3.5: Case inflections and derivational suffixes with cardinal numbers

3.3 Morphology of location words

Location words in Kuku Yalanji belong to a closed word-class with different inflectional properties from those of nominals. A full list of attested location words is:

wangkar	'upwards'
wangkarwangkar	'up there, above'
bada	'downwards'
badabada	'down there, underneath'
walmbi	'left'
jakuy	'right'
kungkarr	'north'
kuwa	'west'
jibarr	'south'
naka	'east'
yuba	'near'
ngundu	'close by'
kalakalbay	'far away'
wuyar	'on top'
ngubar	'on the other side of the river'
yiringkurr	'everywhere'
kudamundu	'behind'
yarra	'over there (general direction)'

68 Chapter 3

Where semantically possible, location words may have temporal extension. Thus yuba and ngundu may be used in the sense of 'near' and 'close by in time', and kudamundu may mean 'afterwards'.

3.3. 1 Locational systems

Three symmetrical locational systems are discernable in the above list of location words: up/down, left/right, and compass points. These and the general direction marker yarra will be discussed in the following subsections.

(i) Wangkar 'up', bada 'down'

With respect to the surroundings of the settlement at W ujal W ujal on the Bloomfield River wangkar refers to 'up the river' and bada to 'down the river'. When indicating direction 'up a mountain' or a downward path other than along the river valley, the general direction marker yarra together with an appropriate gesture is used in connection with wangkar or bada. In an extended sense wangkar refers to a place or direction 'above speaker (in the air, in a tree, etc.)' and bada to a place 'below speaker (on the ground, under water, etc.)'.

(ii) Walmbi 'left', jakuy 'right'

These terms have only limited use in Kuku Yalanji, applying apparently only to body parts. They generally occur as reduplicated comitative stems, for example:

jakuy-jakuy-ji	'(my) right side'
jina walmbi-walmbi-ji	'left foot'

By extension walmbi-walmbi-ji refers to a left-handed person.5

(iii) Compass points

Compass points play a most important part in orientation in Kuku Yalanji. As was mentioned above, the concepts left and right refer to the human body and are not used to indicate direction or location in the environment; this is done entirely by compass points and location words 'up/down'. In other words, a Kuku Yalanji speaker does not take his bearings subjectively with reference to his own person, but rather in relation to the objective and absolute concepts of north, east, south, and west. Where an English speaker would give directions as "turn left and then right", a Kuku Yalanji speaker would say, for instance, "tum west and then north". Similarly, the sides of a house are not referred to as "left" and "right" from the perspective of the speaker (which in English is often confusing for the listener), but a house has a "north side", an "east side", etc. according to its position in the actual environment. The following incident illustrates that Kuku Yalanji speakers always take their bearings from the environment. A woman was asked by her son to start the car while he was

⁵ H. Hershberger and R. Hershberger (1 982) list common nouns for 'right hand' and 'right hand side', jurrubiji and malabayi respectively, but do not include a location term for 'right'. Jakuji (jakuy 'right' in my data) is glossed by them as 'left hand', but they concur with my data for location word walmbi 'left' and walmbiwalmbiji 'left-handed'.

trying to push it out of a bog. Not being a driver herself, she asked which pedal to press. She said: Pedal wanyu? Yarra naka, yarra kuwa? 'Which pedal? The eastern or the western one?'

An interesting aspect in the use of compass points in Kuku Yalanji is that they apparently only refer to places on land, but not the sea. Direction towards the sea, roughly east of Kuku Yalanji country, is described as simply jalun-bu 'sea-LOC" or dingal 'base (of tree), end', or also naka-dingal 'south-eastern end (of the land)'. Thus the compass points in Kuku Yalanji do not correspond exactly to a navigational compass, but indicate the following 'westwardshifted' directions:

kungkarr	'north/northwest' (e.g. Cedar Bay, Cooktown, Rossville)
kuwa	'northwest/west' (e.g. Hopevale, Laura)
jibarr	'southwest/south' (e.g. Mount Carbine, Mareeba)
naka	'southeast' (e.g. Mount Pieter Botte, Cairns)

There are cognates in Guugu Yimidhirr and Yidiny corresponding to the general directions indicated by the Kuku Yalanji terms. However, in Guugu Yirnidhirr we can distinguish between roots, gungga-, guwa-, dyiba-, naka-, and five sets of locative/allative forms varying in reference to relative distance; e.g. a place a long way to the North would be gunggaalu, a place just to the North is gunggaarra, etc. The Kuku Yalanji terms, which do not make these distinctions, resemble the stem forms that are used in derived inchoative verbs in Guugu Yimidhirr: gunggaarr=mal, guwa=mal, dyibaarr=mal, naga=mal. Djabugay has cognates gunggarri 'north', djiwarri 'south', guwa 'west' (but djilnggu for 'east').

(iv) General direction yarra

Yarra expresses no specific directional reference in itself. It may occur with other location words or nouns in locative case and implies 'over there in the direction of X'. When used on its own it is always accompanied by a gesture of hand or head towards the direction referred to (see also deictic yarra 'there, yonder' in Guugu Yimidhirr, and suffix -djirra 'towards' in Djabugay).

3.3.2 Location words - inflections

Location words occur only in local cases. They differ formally from nouns in that the root itself implies locative (with the function of 'location at' or 'direction towards') and therefore does not take locative case inflection. Perlative 'through' apparently cannot be used with location words since they do not define a space that can be traversed, but refer to a more general location.

Wangkar 'up', and bada 'down' distinguish between 'location' and 'direction' in their rootor stem-form. The simple root implies direction, e.g. bada 'downwards', and the reduplicated stem implies location, e.g. badabada 'down there, underneath'. Both simple and reduplicated forms may take further local inflections.

The inflectional system and forms for location words are:

locative	∅
ablative	-muny	(=‘neutral’ nominal inflection)
abessive	-mundu	(= nominal inflection)

70 Chapter 3

Some examples of location words in different cases are:6

• (1 06) Buliman-angka ngayku nganjan naka police-ERG:pt(A) 1 sg.POSS.ABS(O) f ather.ABS(O) east.LOC 'The police took my father "Cairns-way".' (i.e. south-east) wundi-ny. take-PAST
• ( 107) Mary ngundu jana-y Queenie-nda. Mary .ABS(S) close. by .LOC stand-NONP AST Queenie-LOC:pt 'Mary is standing close to Queenie. '
• (1 08) Nyulu kada-ny ngubar-muny. 3sg.NOM(S) come-PAST other.side.of.river-ABL 'He came from the other side of the river. '
• ( 1 09) N gayu wulngku kalakalbay-mundu I sg.NOM(A) song.ABS(O) far.awaY-ABESS nyaji-l. hear-NONPAST 'I hear a song from a long way away.'
• ( 1 1 0) Bama yirrka-n-yirrka-y badabada-mundu. Aborig.ABS(S) shout-REDUP-NONPAST down.there-ABESS 'People are singing out from down there.'

The location word kudamundu 'behind' could be a fossilised abessive form, although kuda cannot be analysed in this context. (The form -kuda is otherwise used as temporal clitic, meaning something like 'meanwhile'; see §3.l0.1 and §5.2.2.) Note that kudamundu can take ablative but not abessive inflection, for example:

• ( 1 1 1) Wulbuman-angka nyunguny wukurri-ny kudamundu(*-mundu). old.woman-ERG:pt(A) 3sg.ACC(O) follow-PAST behind.LOC(*ABESS) 'The old woman followed him. ' OR ' ... followed from behind.'
• (1 1 2) Warru jalama-ny juku-muny kudamundu-muny. yg.man.ABS(S) jump-PAST tree-ABL behind-ABL 'The young man jumped [out] from behind the tree.'

3.4 Morphology of time words

Words with temporal reference belong to a separate word class in Kuku Yalanji that is distinguished from both nouns and location words by its morphology. Also, time words do not co-occur with nominals but constitute phrases in their own right. The members of this word class are:

nyiku	'now, today'
yilaybaja	'later on today'
yilay-yilay	'late afternoon'
wunkuny	'next day, tomorrow'
juma	'eventually, soon, next time'
yilayku	'yesterday'
ngadiy-	'a long time'

Locative, which has zero inflection, is glossed in this section only, but not in any further examples or texts.

70 Chapter

The root ngadi(y)- ⁷occurs only in inflected form, ngadi(y)-ngka(DAT) 'for a long time' and ngadi(y)-muny (ABL)’ from way back’, or with clitic -ku, i.e. ngadi(y)-ku ‘a long time ago'. (Clitic -ku might have perfective implication with time words, e.g. yilayku 'yesterday'.) The form yilay which occurs in three time words is another possible, but not analysable, root that cannot occur in isolation.

Time words in simple root form without inflection refer to 'in, at, during a given time', which is similar to the locative implication of uninflected location words, for example:

• ( 1 1 3) Yilayku nyulu bambay-ma-ny. yesterday 3sg.NOM(S) sick-INCHO-PAST 'He became ill yesterday. '
• ( 1 1 4) Wunkuny nyulu dunga-y doctor tomorrow 3sg.NOM(A) go-NONPAST doctor.ABS(O) see-PURP 'Tomorrow he will go to see the doctor.' nyaji-nka.

Apart from the root form, time words may occur in dative and ablative case, taking regular 'neutral' inflections. Dative implies 'for a given time' and ablative 'since, from a given time', for example:

(115)	Ngayu	mayi	kanbal	kuji-l	yilaybaja-ka
	1sg.NOM(A)	food.ABS(O)	some.ABS(O)	keep-NONPAST	later-DAT
'I keep some food for later on today.'

( 1 1 6) Nyulu kambi mukul kuji-l-kuji-ny ngadi-muny. 3sg.NOM(A) clothes.ABS(O) old.ABS(O) keep-l-REDUP-PAST long.time-ABL 'He has been keeping some old clothes from a long time [ago].'

Juma 'soon' has two alternative forms in dative case which are used in free variation. These arejuma-ka andjumay-ngka, which may indicate that there is a suppletive or archaic rootjumay, for example:

(1 1 7) Nyulu kulji mani-ny, yijarri-ny jumay-ngkal-ka. 3sg.NOM(A) rock.ABS(O) get-PAST put-PAST nexttime-DAT 'He got a rock and set it up for the next time. ' (text 1 2, line 60)

Other time stretches are referred to by common nouns or stems that take regular nominal inflections, for example:

wujurr	'night, darkness'
wujurr-bu (LOC)	'at, during the night'8
warngku	'one day and night' ('sleep')
kija	'moon, month'
bulur-iji	'cold season' ('dew/chill -COMIT')
wungar-iji	'hot season' ('sun-COMIT')

1 We can assume that this root has final -y, which is deleted after vowel i- (see §2.5.2), since it takes the dative allomorph for y-final roots. However, because this root never occurs in uninflected form, I do not represent root-final y in further examples.

8 The beginning of the Christmas carol Silent Night has been translated as wujurrbu ngadiku ‘during the night a long time ago'.

72 Chapter 3

Times of the day may be referred to by descriptive phrases, e.g. wungar dakal 'sunrise' ('sun climb'), wungar wangkarwangkar 'mid-morning to early afternoon' ('sun above').

3.5 Personal pronouns

Kuku Yalanji has free pronoun forms which are used in reference to animate beings, particularly humans. Personal pronouns distinguish three numbers - singular, dual and plural - as well as inclusive and exclusive for first person non-singular. However, the distinction between inclusive and exclusive, particularly in the dual, is not always made. Where this is the case, the exclusive form is used rather than the inclusive form. Thus it appears that the exclusive forms are taking on an unmarked meaning, and inclusive forms may become obsolete.9

Second person plural is used as an honorific address term towards some in-law relations, particularly brother-in-law.

3.5. 1 Paradigm

Core-syntactic case marking for personal pronouns follows a nominative/accusative pattern with an unmarked form for both transitive and intransitive subject (nominative) and a marked form for transitive object (accusative). This is of course different from nominal syntactic case marking where intransitive subject and transitive object take zero inflection and only the transitive subject has a marked form. Except for this difference in the core-syntactic case-marking system, which is widespread in Australian languages, pronouns are used in all cases that have a 'potent' inflectional allomorph, i.e. dative, general locative and ablative.

Semantic constraints do not allow pronouns to occur in instrumental or perlative case. Instrumental inherently implies inanimacy, which is in contrast to the animate reference of pronouns. R. Hershberger ( 1 964c:60) attested instrumental pronoun forms based on the possessive stem as in the following example (quoted from Hershberger):

(1 1 8) Ngayu jana-nda-mundu Carol kunin. 1sg.NOM(A) 3pl-POSS-INST Carol.ABS(O) hit-PAST 'I hit Carol with them'

Such sentences were rejected by my informants, who replaced the pronoun with a demonstrative which can have either animate or inanimate reference. Similarly, perlative is not used with pronouns since animate beings are not regarded as something that can be 'passed through'. Construed sentences like 'the spear went right through him' proved unacceptable and were rendered as 'the spear got stuck in him'. Abessive pronoun forms were rejected by informants in favour of ablative forms.

⁹ Djabugay does not distinguish between inclusive and exclusive, nor are there dual forms in Djabugay. The first person plural form, cognate with the Kuku Yalanji exclusive form, is: nganyji (nominative), nganyjiny (accusative) (Hale 1 976b:237; Patz 1 99 1 :274).

A full paradigm of inflected pronoun forms and comitative and possessive sterns with zero inflection is given in Table 3.6. This paradigm can be analysed as follows. For nonsingular pronouns the nominative form serves as root for all other cases. The accusative suffix is:

-ngan / V- -an / C-	in Yalanji dialect
-niny / V- -iny / C-	in Nyungkul dialect

Other suffixes follow the nominal case inflection paradigm. Note that the possessive suffix does not show a dialect difference between Yalanji and Nyungkul as it does with nouns. The Nyungkul nominal possessive suffix -(V)ngV is used for non-singular possessive pronouns in both Yalanji and Nyungkul.

For second and third person singular there are two distinct forms for nominative and for possessive with the possessive form serving as base for the other cases. The accusative suffix IS:

-ny	in Yalanji
-niny	in Nyungkul (the same as for non-singular forms)

First person singular shows three distinct forms, which are the same in both dialects:

nominative:	ngayu
accusative:	nganya
possessive:	ngayku

The possessive form is used as base for all other case inflections.

To summarise, we can say that there is a two-way split in the Kuku Yalanji pronoun paradigm according to base forms. For non-singular pronouns the nominative form serves as base throughout the paradigm. Singular pronouns have distinct forms for nominative and possessive with all other cases being built on the possessive form; only first person singular has a suppletive form for accusative. Table 3.7 shows pronominal base forms and their suffixes.

3.5.2 Possessive stems

Where a possessive pronoun modifies a noun as part of a wider noun phrase, the pronoun normally agrees in case with the head noun, but does not necessarily have to do so: an uninflected possessive stern may be used. If the head noun is in absolutive case, the accompanying possessive pronoun also remains uninflected, for example:

(119)	Ngayu	nyungu	yaba	nyaji-ny.
	1sg.NOM(A)	3sg.POSS.ABS(O)	eB.ABS(O)	see-PAST
	'I saw his elder brother.'

(120) Ngayku kaya-ngka 1sg.POSS dog-ERG:pt(A) child.ABS(O) bite-PAST 'My dog bit a child. ' karrkay bayka-ny.

..

	Nominative		Accusative	Dative	Locative	Ablative	Comitative +zero inflection	Possessive +zero inflection
		Yalanji	Nyungkul
Sg. 1. 2. 3.	ngayu yundu nyulu (yulu)*	nganya yununy nyunguny	nganya yununiny nyunguniny	ngaykunku yununku nyungunku	ngaykundu yunundu nyungundu	ngaykundumuny yunundumuny nyungundumuny	ngaykunji yununji nyungunji	ngayku yunu nyunju
Du.1.exc. inc. 2. 3.	ngali ngaliny yubal bula	ngalingan ngalinyan yubalan bulangan	ngalininy ngalinyiny yubaliny bulaniny	ngalinka ngalinyanka yubalanka bulanka	ngalinda ngalinyanda yubalanda bulanda	ngalindamuny ngalinyandamuny yubalandamuny bulandamuny	ngalinji ngalinyanji yubalanji bulanji	ngalinga ngalinyanga yubalanga bulanga
Pl.1.exc. inc. 2. 3.	nganjin ngana yurra iana	nganjinan nganangan yurrangan ianangan	nganjininy ngananiny yurraniny iananiny	nganjinanka ngananka yurranka iananka	nganjinanda ngananda yurranda iananda	nganjinandamuny nganandamuny yurrandamuny ianandamuny	nganjinanji ngananji yurranji iananji	nganjinanga ngananga yurranga iananga

Table 3.6: Personal pronoun paradigm

* The two alternative forms for third person singular are used in free variation, but nyulu is more common than yulu

	Base forms		Inflections
	(Nominative)	(Possessive)	Accusative	Dative	Locative	Ablative	Comitative
Sg.1.	ngayu	ngayku-	Yalanji nganya -ny	Nyungkul nganya
2.	yundu	yunu-
3.	nyulu	nyungu-
Du.1.exc.	ngali-	-(a)nga	-(ng)an	-(n)iny				-( V ) n kV*
inc.	ngaliny-
2.	yubal-
3.	bula-
Pl.1.exc.	nganjin-
inc.	ngana-
2.	yurra-
3.	jana-

Table 3.7: Personal pronoun base forms and inflections

* This and the following inflections have allomorphs with initial vowel following root-final consonants (see nominal case inflections, Table 3.1). Any vowel V is determined by vowel harmony rule (see §2.5.1).

76 Chapter 3

If the possessive pronoun is inflected in agreement with the head noun, a special possessive pronoun stem has to be formed which takes suffixes from the 'neutral' set of case inflections. (Nominal possessive stems inflect in the same way, see §3.2.3.2.) Again, the formation of these stems is different for singular and non-singular forms:

singular pronoun: possessive root + non-singular pronoun: nominative root + -ndVmun/V--wun- ,.., -ngun- -VndVmun/C-

This shows that non-singular pronouns form possessive sterns in the same way as nouns (see §3.2.3.2). A full paradigm of possessive pronoun stems is given in Table 3.8, followed by some examples in sentences.

Sg.	1.	ngaykuwun- *
	2.	yunuwun-
	3.	nyunguwun-
Du.	1.exc.	ngalindamun-
	inc.	ngalinyandamun-	non-zero inflection
	2.	yubalandamun-
	3.	bulandamun-
Pl.	1.exc.	nganjinandamun-
	inc.	nganandamun-
	2.	yurrandamun-
	3.	janandamun-

Table 3.8: Possessive pronoun stems with non-zero inflection

*The stem-forming suffix for singular pronouns has also been recorded as -ngun. The two forms occur in free variation.

• (121) Ngayku-wun-du kaya-ngka karrkay bayka-ny. 1sg-POSS-ERG(A) dog-ERG:pt(A) child.ABS(O) bite-PAST 'My dog bit a child. ' cf. ( 1 20)
• ( 1 22) Ngayu wawu jirray yunu-wun-ku ngamu-nku. Isg.NOM(S) like much 2sg-POSS-DAT mother-DAT:pt 'I like your mother very much.'
• ( 1 23) Yanya mani daya yubal-andamun-bu nganjan-anda. this.ABs(o) money.ABS(O) give.lMP 2du-POSS-LOC father-Loc:pt 'Give this money to your [dual] father!'
• ( 1 24) Buliman kada-y jana-ndamun-muny bayan-muny. police.ABS(S) come-NONPAST 3pl-POSS-ABL house-ABL 'A policeman is coming from their house.'

Like nouns, pronouns do not have special forms to indicate 'possessor of a possessor'. The regular possessive stem is used and there is a preference for locative inflection, regardless of the case of head noun, for example:

(1 25) Nyulu yunu-wun-bulyunu 3sg.NOM(A) 2sg-POSS-LOcJ2sg.POSS.ABS(O)

kangkal-ba mayi wundi-ny. own.child-POSS.ABS(O) food.ABS(O) take-PAST 'He took your child's food. '

(126) N gayku-wun-bul -du babarr-andamun-du kaya-ngka I sg.POSS-LOO-ERG(A) eZ-POSS-ERG(A) dog-ERG:pt(A) ngayku minya ngaki-ny. I sg.POSS.ABS(O) meat.ABS(O) steal-PAST 'My elder sister's dog stole my meat.'

(It must be pointed out that such constructions were only obtained through elicitation.)

3.5.3 Historical notes

From the historical point of view the system and forms of Kuku Yalanji personal pronouns are quite conservative. It may be assumed that pA pronouns had singular/ dual/plural distinction and the eastern languages generally show relatively few changes in this system. As for the actual roots, similar forms can be found in many other Australian languages and Kuku Yalanji personal pronoun roots have undergone sound changes and reinterpretations that are quite common.

Dixon (1980:339-346) has reconstructed monosyllabic roots for singular pronouns in S-function with inflections for A and 0 function:

	S	A	O
1sg	ngay	ngay + DHu	ngay + NHa
2sg	ngin	ngin + du	ngin + NH
3sg	NHu	NHu + lu	NHu + NHa

As a general trend Australian languages at some stage in their development began to abolish monosyllabic roots and in most languages, Kuku Yalanji among them, the pA agentive forms were reinterpreted as roots covering both A and S function. Following common phonological changes (outlined in Dixon 1 980) this yielded in Kuku Yalanji:

1sg	ngayu
2sg	yundu
3sg	nyulu (yulu)

The suppletive first person singular accusative form nganya can be analysed as directly related to the pA 0 form ngay-NHa, and the first person singular possessive root ngayku is indeed identical with the corresponding pA form ngay-ku if the assumption holds true that -ku is the pA genitive suffix. (However, Kuku Yalanji cannot provide any more evidence for GEN -ku than, for instance, Warrgamay, which also has I sg POSS ngayku, but does not use GEN -ku anywhere else in its pronoun paradigm.)

Proto Australian non-singular forms are thought to have had one root for S/A function with suffix -NHa for 0 function. The assumed pA forms are (Dixon 1 980:334-339):

r

1du	ngali
2du	NHu(m)balV
3du	bula
1pl	ngana
2pl	NHurra
3pl	DHaNa

The distinction between inclusive/exclusive first person forms appears to be an innovation. Distribution in modern languages shows that where this distinction is made ngali and ngana usually indicate 'inclusive', while 'exclusive' forms are either based on ngali and ngana or are quite unrelated to these. In Kuku Yalanji the reverse applies for first person dual forms: ngali implies 'exclusive', while the 'inclusive' pronoun involves an increment to this form: ngalim!.. First person plural pronouns, on the other hand, concur with the general Australian pattern in that ngana is the 'inclusive' form. Nganjin 'lpl exc.' could be based on ngana (cf. Guugu Yirnidhirr inland dialect nganhdhaan ' lpl'), but there is no real evidence to show that it is.

The second person non-singular forms yubal and yurra have developed by lenition of the initial consonant: NH > ny > y. (Similar lenition of the first consonant has also applied to second and third person singular.) The final vowel of 2du *NHu(m)baIV has been dropped in Kuku Yalanji. The fact that inflections to yubal involve an initial vowel, e.g. Dative yubal-anka in contrast to ngana-nka, is probably of no consequence since all 'potent' inflections in Kuku Yalanji have allomorphs with initial vowel following root-final consonants.

Third person plural jana shows a regular reflex of pA DH (DHaNa > jana), while third person dual bula involves no change at all.

3.6 Interrogative pronouns

3.6.1 Paradigm and morphology

Interrogative pronoun roots in Kuku Yalanji show the common Australian distinction between 'who', 'what' and 'where'. In many Australian languages interrogative pronouns can have both interrogative and indefinite meaning, e.g. 'who/someone'. This is not so for Kuku Yalanji interrogative forms 'who' and 'what', which have only interrogative sense. 'Where', on the other hand, may imply both 'where' and 'somewhere'. To express 'something indefinite' the generic term junjuy is used in reference to anything inanimate or animate except humans, but there seems to be no way to express 'some indefinite person'. (In precontact times the generic term bama 'Aboriginal person' would presumably have been sufficient for this function.)

	Yalanji	Nyungkul
'who'	wanya	wanju
'what'	wanyu	wanyu
'where'	wanja	wanja

	Table 3.9: Interrogative pronouns in Yalanji and Nyungkul
--	-----------------------------------------------------------	--

The form for 'who' differs in the Yalanji and Nyungkul dialects but 'what' and 'where' are the same (see Table 3 .9). 'Who' and 'what' inflect according to the nominal system of case marking; that is, core syntactic inflections distinguish between transitive subject and intransitive subject/transitive object. (This is different from the most frequent pattern in Australian languages where 'who' inflects in a nominative/accusative pattern like pronouns, and 'what' like nominals in an ergative/absolutive pattern.) 'Who' takes suffixes from the 'potent' set of case inflections, and 'what' takes inflections from the 'neutral' set. 'Where' is used only in local cases; it is also the base for further interrogatives like 'when' and 'how'.

A paradigm of inflected 'who' and 'what' forms and their possessive and comitative stems is given in Table 3 . 1 0. The Yalanji form wanya 'who' shows regular nominal suffixes for all but the possessive stem with zero inflection. This stem, in both Yalanji and Nyungkul, takes a suffix identical to the 'neutral' locative inflection which is also used for possessive with a subset of kinship terms (see §3.2.4). The Nyungkul form wanju 'who' also takes regular nominal suffixes; however, on the evidence from all inflected forms except ergative we have to recognise an underlying root with final y, namely wanjuy. This root takes the regular suffixes with initial vowel that follow a root-final consonant.

	'who'		'what'
	Yalanji	Nyungkul
ABS	wanya	wanju	wanyu
ERG	wanyangka	wanjungku
INST			wanyurri(y)nju (wanyurrimundu)
DAT	wanyanka	wanjuyunku	wanyurri(y)ngku
LOC	wanyanda	wanjuyundu	wanyurri(y)mbu
ABL10	wanyandamuny	wanjuyundumuny	wanyurri(y)muny
COMIT11	wanyanji	wanjuyunji	wanyurri(y)nji
POSS + zero inflection	wanyamba	wanjuymbu
POSS + non-zero inflection	wanyandamun-	wanjuyundumun-

Table 3.10: Paradigm of interrogative pronouns 'who' and 'what'

Wanyu 'what' shows an underlying stem wa nyu rri(y )- for all but absolutive case, followed by regular case inflections from the 'neutral' set. For ergative/instrumental an alternative form wanyurri-mun-du may be used. Speakers who use this latter form seem to have reinterpreted the stem as ending in i and add the catalytic suffix -mun- onto the trisyllabic

¹¹ Comitative stems with non-zero inflections beginning with nasal m or a stop take the comitative allomorph -jirr (see Table 3.3).

¹⁰ These forms are not favoured for the local sense of ablative. Preference is given to wanjamuny 'where from'.

80 Chapter 3

root. Speakers' intuitions about the inflected forms of wanyu appear to be uncertain in another respect as well. Note that vowel harmony does not match the suffix vowel with the final stem vowel, e.g. *wanyurri.(y)njQ, but with the final vowel of the uninflected root, e.g. wanyurri(y)nju. This might suggest that the whole of -rri(y) + inflection is in fact interpreted as the suffix.

The morpheme -rri(y)- may have been a stem-forming suffix at some stage. It also seems to be present in wanjarrinya 'how many' (see below), in the plural demonstrative forms in Kuku Yalanji and Guugu Yimidhirr, e.g. yinyarriny 'those people' (see §3.7), and in Yidiny yinga-rriny 'this kind' and wanyja-rriny 'what kind', where this morpheme cannot be analysed either (see Dixon 1 977:1 97-1 98).

Wanyu has no possessive forms, since inanimate things are considered incapable of owning anything. The local term wanjamuny 'where from' is used to express 'what does that belong to'.

The dative form of 'what', wanyurringku,12 is used in the sense of 'why' (lit. for what [purpose]). However, if the question contains a reference to a location, e.g. 'why did you go there', the locative form of wanyu may be used, although the dative form is possible, for example:

(127) a.	Wanyurrimbu	yundu	kaya	wundi-ny	bayan-ba?
	what.LOC	2sg.NOM(A)	dog.ABS(O)	take-PAST	house-LOC
					'Why did you take the dog into the house?' (and not somewhere else)

b. Wanyurringku yundu kaya wundi-ny bayan-ba, what.DAT 2sg.NOM(A) dog.ABS(O) take-PAST house-LOC ngalkal-ba-ku bawa-nyaku. outside-LOC-EMPHI leave-IRR 'What did you take the dog into the house for, you should have left it outside.'

The locative and dative forms of wanyu imply a different focus of the question. The locative form enquires after the reason for a particular choice of location as in (1 27a): 'why the house and not some other place?'. The dative form, on the other hand, asks for the reason that prompted a particular action as indicated by the translation for ( l 27b).

A specific use of the comitative form has to be noted: it is always used when enquiring about the nature of an illness, for example:

(1 28) Yundu wanyurrinji bambay? 2sg:NOM(S) what.COMIT sick 'What are you ill with?'

A commonly used prompt for further information in a conversation involves wanyu and suffix -baja 'again, Degree' (see §3.2.3.8 (ii) and §3.10.3): wanyubaja 'what else'.

Wanja 'where, somewhere' can be classed among the location words and as such implies general locative by its uninflected root and takes only local cases (cf. location words §3.3 .2). But it seems that the uninflected form is being replaced by another form: wanjabu. H. Hershberger and R. Hershberger (pers. comm.) reported that several years ago forms wanja and wanjabu were used in free variation. But it appeared that wanja was the more general term implying 'where at/where to' while wanjabu asked more specifically for location

¹² Stem-final y, which conditions the inflectional form but is lost after i- (see §2.5.2), will not be represented in further examples.

'where at'. My data includes only two occurrences of wanja, but copious examples of wanjabu in both locative and allative sense, for example:

• ( 1 29) Yunu ngamu wanjabu? 2sg.POSS.ABS(S) mother.ABS(S) where 'Where is your mother?'
• ( 1 30) Yundu wanjabu dunga-y? 2sg.NOM(S) where go-NONPAST 'Where are you going?'

Sentence ( 1 30), or simply yundu wanjabu is the common form of greeting people. So it seems that wanja is being wholly replaced by wanjabu. The etymology of wanjabu is obscure. It may possibly have developed from a phrase wanja bubu 'where place'. The paradigm of wanja forms is set out in Table 3.1 1.

general
Locative		wanjabu
		( wanja )
Ablative		wanjamuny
Abessive		wanjamundu

Table 3.1 1: The forms of wanja 'where'

As mentioned above, wanja(bu) can have both interrogative and indefinite meaning. Consider:

( 1 3 1) Yunu scissors wundi! Ngayku 2sg.POSS.ABS(O) scissors.ABS(O) take.IMP I sg.POSS.ABS(S) scissors wanjabu. scissors.ABS(S) somewhere 'Bring (take) your scissors ! My scissors are somewhere (1 don't know where I put them). '

More examples of the indefinite use of wanja forms can be found in text 51, lines 2 and 54, text 1 2, line 2.

Other interrogative roots are based on wanja, which invites the interpretation of wanja as 'general interrogative' root. Such forms are:

wanjawanja	'when'
wanjarr	'how'
wanjarrinya	'how many'
wanjinya or	'which one'
wanjaku

Wanjarr 'how' usually takes the adverbial suffix -ku (see §3.8.6.2), but may occur without it, especially with intransitive verbs, for example:

82 Chapter 3

• ( 1 32) Yundu wanjarr kada-ny-baja? 2sg.NOM(S) how come-PAST-again 'How did you come back?'
• ( 1 33) Yurra wanjarr(-ku) minya 2pl.NOM(A) how (-ADV) meat.ABS(O) 'How do you cook freshwater eel?' kubarr waju-l? freshwater .eel.ABS(O) cook -NONP AST

Wanjinya and wanjaku 'which one' seem to be freely interchangeable, for example:

( 1 34) Wanjinya/wanjaku yunu bayan? which. one 2sg,POSS.ABS house.ABS 'Which one is your house?'

A stem-forming suffix -nguy can be used with either wanyu 'what' or wanja 'where, general interrogative' to imply 'what kind', for example:

(135)	Kuyu	wanyunguy	jana	mani-ny?
	fish.ABS(O)	what.kind.ABS(O)	3pl.NOM(A)	get-PAST
	'What kind of fish did they get?'
(136)	Jana	walu	wanyunguy?
	3pl.NOM(S)	face.ABS	what.kind-ABS
	'What kind of people are they?'
(137)	Yundu	wanjanguy-ngku	kambi-ka	wawu, yinya-ka
	2sg.NOM(S)	what.kind-DAT	dress-DAT	want/like that-DAT
		nganka-nganka-jirr-ka?
		flower-REDUP-COMIT-DAT
	'Which kind of dress do you like, the one with the flower pattern?'

3. 6.2 Historical notes

It appears that all interrogative forms in Kuku Yalanji can be traced back to a general monosyllabic interrogative pA root *waNH- (see Dixon 1 980:374-378) and inflected forms of this root have been reinterpreted as disyllabic roots (cf. reinterpretation of singular pronouns, §3.5.3). Interestingly, Yalanji and Nyungkul have reinterpreted the form for 'who' in different ways. In Nyungkul the ergative form *waNH-du seems to be the underlying form for wanju, whereas Yalanji has simply added vowel a to the original absolutive form, i.e. wanya, in the same way as Yidiny has reinterpreted this root (see Dixon 1 980:374-375). The interrogative wanyu 'what' may also have been derived from the historic ergative form with further reduction of the intermorphemic consonant cluster: NHd > nyj > ny (cf. wanhu 'who' in Guugu Yimidhirr, Haviland 1 979a:69). Wanja 'where' is presumably based on the pA locative form *waNH-da. Note that -nj- in wanju and wanja are phonetically homorganic clusters according to the assimilation rule n-> nj / -j, but they are not phonemically homorganic, since the preceding vowel does not have a palatal offglide, e.g. wanjja and not wainyja. (For assimilation rules see §2.4.)

Kuku Yalanji interrogative pronouns resemble most closely those of Yidiny (see Dixon 1 977a:1 87). Guugu Yimidhirr 'who' and 'where' are also based on pA *waNH, but 'what' uses the root ngaan- (Haviland 1 979a:69), which is found as the base for 'who' or 'what' in Warrgamay (Queensland), Ngiyambaa (New South Wales) and some Victorian and Western Desert languages among others (see Dixon 1 980:373). Djabugay, on the other hand, has monosyllabic interrogative forms with long final vowel which are cognate to the second syllables, i.e. not the pA roots, of the Yidiny and/or Kuku Yalanji forms (see Hale 1 976b:238; Patz 1 99 1 :277-278).

	Djabugay	Yidiny	Kuku Yalanji
'who'	dju:	wanyju	wanju (Nyungkul dialect)
'what'	nyi:	wanyi	wanyu
'where'	dja:	wanyja	wanja

That is to say that Djabugay has undergone more changes, reverting to monosyllabic roots based on the increment to the pA roots.

3.7 Demonstratives

Kuku Yalanji has two demonstrative pronouns and two demonstrative location words, all of which follow the distinction between 'thislhere' and 'that/there'. The forms of 'this' and 'here' differ in the Yalanji and Nyungkul dialects as shown in Table 3 . 1 2.

	Yalanji	Nyungkul
'this/these'	yanyu	yanya
'that/those'	yinya	yinya
'here'	yaluy	yalay
'there'	yinyay	yinyay

Table 3.12: Demonstratives in Yalanji and Nyungkul

Demonstrative pronouns imply singular or plural according to context, and may refer to animate or inanimate beings. They have the same case-marking possibilities as nomina Is , whereas demonstrative location words occur only in local cases. The paradigm of inflected forms and possessive and comitative stems in Table 3. 13 shows that demonstrative pronouns have suppletive bases for all inflected forms and stems which are either identical to the local demonstrative forms, e.g. in the dative forms and Nyungkul yalaymba 'this-LOC/POSS', or have dropped the final y of the local demonstratives as in the other forms. (Note that the same possessive suffix is used that occurs with interrogatives and some kinship terms.) The Yalanji forms for 'this' also show variation in the second base voweL In the ablative, abessive and comitative forms this is a, whereas all other forms have second vowel u. In the Yalanji dialect the 'neutral' locative form of 'this', yaluymbu, is often rejected and preference given to the corresponding local demonstrative, yaluy.

84 Chapter 3

	'this'		'that'
	Yalanji	Nyungkul
ABS	yanyu	yanya	yinya
ERG/INST	yalubu	yalabu	yinyabu
ERG:pt	yalungku	yalangka	yinyangka
DAT	yaluyngku	yalayngka	yinyayngka
DAT:pt	yaluyunku	yalayanka	yinyayanka
LOC	(yaluymbu)	yalaymba	yinyamba
LOC:pt	yaluyundu	yalayanda	yinyayanda
ABL	yalamuny	yalamuny	yinyamuny
ABESS	yalamundu	yalamundu	yinyamundu
PERL		yalaburr	yinyaburr (Nyungkul)
COMIT	yalanji	yalanji	yinyanji
POSS + zero inflection	yalumbu	yalaymba	yinyamba
POSS + non-zero inflection	yaluyundumun-	yalayandamun-	yinyandamun-

Table 3.13: Paradigm of demonstrative pronouns

Yinya 'that' and the suppletive forms of 'this' can form plural demonstrative pronouns with suffix -rriny (cf. suppletive root wanyurri(y)- 'what' in §3 .6. 1 ). These forms are used only in reference to human beings and apparently can only occur in absolutive case. They are:

	Yalanji	Nyungkul
'these people'	yalurriny	yalarriny
'those people'	yinyarriny	yinyarriny

The paradigm of local demonstratives in Table 3.14 shows further similarities with the inflected forms of demonstrative pronouns. It can be seen from this table that the Yalanji demonstrative 'here' behaves exactly like location words. While the uninflected root implies general locative, the ablative and abessive forms show regular inflections. For 'there' and Nyungkul 'here', on the other hand, either the root or an inflected locative form may be used. (In Nyungkul the inflected form yalaymba is preferred.) The perlative forms in the Nyungkul dialect show an idiosyncratic suffix -ngkarr which does not occur with nominals. The ablative and abessive forms for 'there' and Nyungkul 'here' are identical to the corresponding demonstrative pronoun forms.

	'here'		'there'
	Yalanji	Nyungkul
LOC	yaluy	(yalay)	yinyay
		yalaymba	yinyaymba
ABL	yaluymuny	yalamuny	yinyamuny
ABESS	yaluymundu	yalamundu	yinyamundu
PERL	yalangkarr	yalangkarr	yinyangkarr (Nyungkul)

Table 3.14: Paradigm of local demonstratives

The conclusion that may be drawn from these morphological details is that there is no clear distinction between demonstrative pronouns and local demonstratives in the inflected forms.

Both Guugu Yimidhirr and Djabugay have demonstrative systems that do not distinguish between pronominal reference and local reference (Haviland 1 979a:72-74; Patz 1 99 1 :273ff.). In Djabugay demonstratives 'thislhere' and 'that/there' show singular/plural distinction in their root forms and function as third person singular and plural pronouns.

'Here' and 'there' in Kuku Yalanji may combine with the stem-forming suffix -arra 'inhabitants' :

Yalanji	yaluyarra	'the people of this place'
Nyungkul	yalayarra
	yinyayarra	'the people of that place'

Although demonstrative pronouns and local demonstratives are not always different in form, they are distinct in function. Demonstrative pronouns, similar to personal pronouns, can replace or co-occur with a noun, for example:

(138)	Yalu-ngku jalbu-ngku	nganya	binal-bunga-ny
	this-ERG:pt(A) woman-ERG:pt(A)	1sg.ACC(O)	know-CAUS-PAST
	marra-ka.
	Zamia.nut-DAT
	'This woman taught me about zamia nuts.'

• ( 1 39) Jana nubi-I-nubi-ny yinyay-anka karrkay-anka. 3pl.NOM(S) search-REDUP-PAST that-DAT:pt child-DAT:pt 'They were searching for that child.'
• (1 40) Yinyay-anda kaban daya. that-Loc:pt letter.ABS(O) give.lMP 'Give the letter to that [person].'

Local demonstratives, on the other hand, are not part of a wider noun phrase but occur on their own, for example:

(141) Nyulu yaluy-muny wanarri-y. 3sg.NOM(S) here-ABL run.REDUP-NONPAST 'He is running from here. '

86 Chapter 3

(1 42) Ngayu jarruka wubul warrmba-bunga-ny I sg.NOM(A) scrubhen.ABS(O) many.ABS(O) uncovered-CAUS-PAST yinyay-mba. there-LOC 'I found a lot of scrubhen [eggs] there. '

Sentences like 'he sat on that rock' are paraphrased as

(143) Nyulu yinyay-mba bundanda-ny, 3sg.NOM(S) there-LOC sit.REDUP-PAST 'He sat there, on the rock.' kulji-nga. rock-LOC

where kuljinga is clearly offset from the clause by a pause and falling intonation on the verb bundandany; that is, kuljinga could be regarded as outside the main clause. (Yinyaymba kuljinga as one phrase is possible, but the construction in ( 143) is preferred.)

3.8 Morphology of verbs

3.8. 1 Introduction

Kuku Yalanji has only a rather small corpus of unanalysable verb roots (170 of a word list of 1 340 roots), but there are many verbal compounds in which some of these roots appear as second member. Furthermore, causative and inchoative derivational processes, which form verb stems from either nominal or verbal roots, are very productive indeed. This feature of a small verb root corpus, in which Kuku Yalanji resembles its northern neighbour Guugu Yimidhirr, is quite unusual for an eastern Australian language, but is widespread in languages of the West.

Kuku Yalanji verb roots refer almost exclusively to 'observable' actions, that is actions that can be seen or heard such as hit, eat, run, speak, call out, etc. Only very few abstract concepts are expressed by verb roots, e.g. nyuyal 'blame', bayjal 'ignore', while many others can be expressed through verbal derivations, e.g. wawurrwawurr-ma-l 'decide, be happy (lit: breathlspirit-COMIT-INCHO)'; see §3.9.1 (a) for information on this stem); through compounds, e.g. milka-jana-y 'listen (lit: ear-stand)" milka-wula-y 'forget (lit. ear-die)'; or through other paraphrases, e.g. milka-bu bayka-l 'think (lit. bite with ear)'. Detailed semantic differentiation accounts for a considerable number of roots, as for instance in the semantic field of 'movement along a path': yimba-l 'go straight through', warrki-l 'turn off at crossroads', barrma-l 'accompany someone for a short distance', nyunba-y 'turn back'.

3.8.2 Verbal word structure

3.8.2. 1 General

Kuku Yalanji verb roots are typically disyllabic. There is a small proportion of trisyllabic roots which will be discussed in the next subsection. There are also four roots with four syllables which are presumably derivations or compounds historically, but which cannot be analysed. These are:

jakalamba-l	'gather firewood, set up fire'
wubulamba-l	'collect, gather' (< wubul 'many'?)
wukaramba-l	'repair'
karangaji-y	'sneak up, stalk'

All verb roots can be recognised to end in a vowel, followed by a conjugation marker, which corresponds to the non-past tense inflection. This conjugation marker is deleted when inflections or stem-forming suffixes are added.

Final-root vowels are a and i with only one occurrence of final u in waju-l 'cook, burn' which has a cognate in Yidiny: wadu-l. The occurrence rate of final a is almost 20% higher than that of final i.

A verb root or compound stem always occurs with an inflectional suffix and can take up to two derivational suffixes between root or stem and inflection. Up to three enclitics may follow the inflected form.

Reduplication normally affects the whole root; only a small number of verbs show partial reduplication of either the first or the second syllable. When a root is reduplicated as a whole the two parts are linked by a filler morpheme whose form depends on the conjugation class of the root or, if a derivational suffix is added, on the conjugation class of the derived stem. In compound stems it is only the second member which is reduplicated.

3.8.2.2 Trisyllabic roots

There are 28 trisyllabic roots, which may be divided into four groups according to their last syllable: -rri, -ma, -ba and others. Some verbs are included in this group which are likely to be regularly formed derivations whose roots have ceased to function independently in Kuku Yalanji. (Inseparable reflexive and reciprocal stems do not fall into this category; they are discussed in their respective sub-sections.) All trisyllabic roots but one belong to the lconjugation class but transitivity varies, although the majority is transitive.

(i) Final syllable -rri

There are 18 attested roots in this group, some of which have cognates among the r-conjugation verbs in Guugu Yimidhirr. These roots have been formed through a reinterpretation process outlined in §3.8.3.2 (ii).

At least three of the roots that do not have cognates among the r-conjugation verbs in Guugu Yirnidhirr seem to be based directly on a nominal root. These are:

bijarri-l	'to dream'	bijarr	'dream'
ngangkurri-l	'to bark'	ngangkurr	'sound of barking'
nganjarri-l	'not know someone'	(Guugu Yimidhirr nganhdha- nganhdha 'foreign person')

(ii) Final syllable -ma

The three verbs ill this group are possibly formed with the inchoative suffix -ma-l. They are:

jalama-l	"jump"
jirayma-l	"long for, crave" (cf. Djabugay jiran "thirst, hunger")
yalama-l	"say, do"

The first two are intransitive, which corresponds to the transitivity class of inchoative stems; but yalama-l 'say, do', which may possibly be based on the suppletive demonstrative form yala 'this', is a transitive verb. 13

(iii) Final syllable -ba

Kadaba-l 'break (itr)" and possibly also three of the four roots with four syllables listed in the previous subsection belong to this group. The three roots with four syllables are all transitive, and final -ba may be a regular reduction of causative -bunga-l (see §3.8.5.1 (i» , e.g. wubulamba-l 'gather' could be analysed as wubul 'many' + causative. However, the cluster mb at the assumed morpheme boundary is quite unexplainable. The intransitive verb kadaba-l has an identical cognate in Guugu Yirnidhirr where Haviland ( 1 979a:80) interprets the final -bal as a nonproductive verbaliser.

(iv) The remaining trisyllabic roots are quite opaque, although they show some similarities in their final syllables:

buraka-l	'hang up to dry (tr)'
yiringka-l	'drag (tr)'
buyandi-l	'put in container (tr)'
dakandi-l	'knock down (tr)'
kalkandi-l	'tighten up, wedge in (tr)
warringa-l	'keep under observation (tr)'

3.B.3 Transitivity and conjugations

3.8.3. 1 Correlation of transitivity and conjugation

As in other Australian languages there is a strict division between transItive and intransitive verbs in Kuku Yalanji. The most basic criterion for distinguishing these grammatical classes is the case inflection on the subject noun occurring with a verb: only a transitive verb can co-occur with a noun in Ergative case while an intransitive verb requires a subject noun in Absolutive case.

Of the 1 70 verb roots in Kuku Yalanji 1 32 (77.6%) are transitive and 38 (22.4%) are intransitive. Six fixed reflexive and two reciprocal stems can be added to the latter group. These stems, although analysable, show considerable extension of the semantic content of the root component and are therefore best regarded as independent lexemes but not necessarily as roots (see §3.8.5.4 (i) and (ii» .

¹³ Compare Dyirbal yala 'this', yalama-l 'do like this, say, tell'. Similar verb forms 'do/say like this/that' exist in many other Australian languages as well as in Central Australian Creole: lakijat < 'like that' (Koch, pers. cornm.).

Formally, Kuku Yalanji verbs are divided into two open conjugation classes, i-conjugation and y-conjugation, according to the conjugation marker -l or -y that is suffixed to a root in what I interpret as the citation form. Apart from this citation form which coincides with the non-past form of a verb the two conjugations differ only in their imperative inflection on disyllabic roots and in their reduplicated forms.

There is an obvious correlation between grammatical class and formal class: y-conjugation verbs are exclusively intransitive while i-conjugation verbs are 92% transitive. The number of known intransitive i-conjugation roots is so small that they may be listed in full:

daka-l	'climb, rise'
janji-l	'swim, bathe'
kalji-l	'vomit'
wala-l	'enter'
walngka-l	'hang'
wandi-l	'come out, wake up'
warrki-l	'turn off at crossroads'
wurrka-l	'ache' (Nyungkul)
kadaba-l	'break'
jalama-l	'jump'
jirayma-l	'crave, long for'

Since the very productively used inchoative verbaliser ma-i belongs to the i-conjugation the overall transitivity proportion of i-conjugation verbs is quite different from that among i-conjugation roots, i.e. the share of intransitive verbs among all i-conjugation verbs is much higher than that among roots only.

A similar proportion of a- and i-final roots is found in both conjugations. Statistical correlation between conjugation, transitivity and final root vowels are shown in Table 3.15.

	l-conjugation	y-conjugation
number of roots	143	27
transitivity	92% transitive	100% intransitive
final root vowel
a	58.5%	74%
i	40.7%	26%
u	0.8% (1 occurrence)	0
number of roots with more than two syllables	27 trisyllabic 3 with four syllables	1 trisyllabic 1 with four syllables

Table 3.15: Relationship between conjugation, transitivity, final root vowel, and number of syllables

3.8.3.2 Development of Kuku Yalanji conjugations

With two open conjugation classes Kuku Yalanji has developed one of the least complex conjugation systems found in Australia. The fact that the two conjugations differ only in two inflections and in reduplicated forms indicates that only minor further simplification will lead to the complete abolition of distinct conjugation classes, a process which has already been completed in some Central Australian languages, as for instance Arandic (Koch, pers. comm.). Languages of the Queensland coastal rainforest region to the South of Kuku Yalanji have simplified their conjugation systems in similar ways. Djabugay (patz 1 99 1 :278ff.) and Dyirbal (Dixon 1 972:54ff.) each have open y- and l-conjugations, but differences between these classes are more distinct than in Kuku Yalanji. Yidiny, located between Djabugay and Dyirbal, has the same open classes plus a small group of rconjugation verbs (Dixon 1 977a:206ff.). Guugu Yirnidhirr (Haviland 1 979a:78ff.), on the other hand, has retained a system of five conjugations and can therefore be considered the most conservative of the languages mentioned above.14 Conjugations in Guugu Yimidhirr are: L- and R-conjugations (both open classes), V-, MA- and NA- conjugations (closed classes with very few members). A comparison of unusual root forms in Kuku Yalanji, i.e. suppletive roots and some trisyllabic forms, with cognates in Guugu Yimidhirr provides some information on how conjugations have contracted in Kuku Yalanji.

(i) Suppletive roots

Three l-conjugation verbs in Kuku Yalanji have suppletive roots that are used as imperative forms; these are:

	Non-past	Imperative
'see'	nyaji-l	nyaka
'get'	mani-l	mana
'give'	daji-l	daya

Nyajil and manil have monosyllabic cognates in Guugu Yirnidhirr MA- and NAconjugations respectively. Listed below are the Guugu Yimidhirr roots with their past tense forms and disyllabic stems that have to be used with purposive inflection:

	Root	PAST	Stem + PURP
'see'	nhaa-	nhaa-dhi	nhaa-dhi-nhu
'get'	maa-	maa-ni	maa-ni-nhu

The disyllabic stems which have the same form as root + PAST are easily recognisable as the Kuku Yalanji roots:

Guugu Yimidhirr	nhaa-dhi >	Kuku Yalanji	nyaji-
	maa-ni- >		mani-

Both transitive roots were reassigned in Kuku Yalanji to the predominantly transitive I-conjugation. (Vowel-length contrast has been lost in Kuku Yalanji.)

The imperative form nyaka 'see!' can be analysed as a mono-syllabic root plus pA imperative suffix *-ga (see Dixon 1 980:4 1 0). (Note that Kuku Yalanji also uses imperative inflection -ka with roots and stems of more than two syllables.) Ngawaygi, a Queensland language further south, has the cognate form nyaa-ga 'see!', and also maa-na 'hold in hand!', cognate with Kuku Yalanji imperative mana 'get!' (Dixon 1 980:399).

¹⁴ Nyawaygi, next but one to the South of Dyirbal, has seven conjugations and many similarities to Guugu Yirnidhirr (Dixon 1 980:397).

No cognates for Kuku Yalanji daji-l 'give' and its suppletive form daya have been determined so far. The development of this form therefore remains unclear.

Another cognate monosyllabic root in Guugu Yimidhirr NA-conjugation supports the pattern of reinterpretation of past tense form or disyllabic stem as root:

	Guugu Yimidhirr	Kuku Yalanji
Root	wu-	wuna-	'lie (down)'
PAST	wu-nay	wuna-ny
'PURP	wu-na-nhu	wuna-nka

The reinterpreted intransitive root wuna- 'lie' has been reassigned in Kuku Yalanji to the exclusively intransitive y-conjugation. 15

(ii) Trisyllabic roots with final syllable -rri

Six of the eighteen attested -rri-final trisyllabic roots in Kuku Yalanji have cognates among the transitive disyllabic members of the R-conjugation in Guugu Yimidhirr. These are:

Kuku Yalanji		Guugu Yimidhirr
janjarri-l	'investigate'	dhunhdha-rr	'ask for'
julurri-l	'wash'	dhulu-rr	'scrub, wash'
nyangarri-l	'pour out'	nhiinggu-rr	'pour out'
walngkurri-l	'ask for'	walnggu-rr	'ask for'
wukurri-l	'follow'	wuku-rr	'follow'
yijarri-l	'put, place'	yidha-rr	'put, place'

Older speakers in Guugu Yimidhirr use past tense inflection -rrinh with R-conjugation verbs (younger speakers use -rrin), e.g. wuku-rrinh 'followed', yidha-rrinh 'placed'. Apparently this past tense inflection has been reinterpreted as a third root syllable -rri- plus general past inflection -ny « -nh) in Kuku Yalanji (see Haviland 1 979a:87). According to their transitivity class the reinterpreted roots were incorporated into l-conjugation.16 Thus the new verb forms in Kuku Yalanji are:

	Citation form	PAST
follow	wukurri-l	wukurri-ny
place, put	yijarri-l	yijarri-ny	and so forth.

Note that only R-conjugation verbs and monosyllabic roots in Guugu Yimidhirr show instances of final root vowel u. In the reinterpretation process outlined above final u ends up in penultimate vowel position which helps to explain why root-final u is extremely rare in Kuku Yalanji verbs.

¹⁵ Haviland (1979a:85) reports that a similar process is taking place in Guugu Yimidhirr where "many younger speakers treat the NA-conjugation verb wu-naa 'lie down' as if it were a regular V-conjugation verb of the form wunaa."

¹⁶ Note that Djabugay has a derivational suffix -rri- which transitivises intransitive verb roots, e.g. warrnki-y 'turn (itr)', warrnki-rri-l 'turn something (tr)' (Hale 1 976b:238; Patz 1 991 :283).

92 Chapter 3

3.8.4 Verbal inflections

3.8. 4. 1 Paradigm

Verbal inflections in Kuku Yalanji include a two-term tense system, nonpast and past, plus imperative, purposive, precautionary, successive, irrealis, and a suffix which has dual function as 'unmarked' inflection (in Nyungkul dialect) and subordination marker in both dialects. Except for nonpast and imperative on disyllabic roots inflectional suffixes are identical for both conjugations. Table 3. 16 shows the paradigm of verbal inflections, while Table 3.17 gives examples of inflected verbs. The latter includes a trisyllabic root to illustrate the imperative suffix that is used with roots or stems of more than two syllables of either conjugation.

	l-conjugation	y-conjugation
NONPAST	-l	-y
PAST	-ny	-ny
IMPerative disyllabic root root/stem with more than two syllables	∅	-y
	-ka	-ka
IRRealis	-nyaku	-nyaku
PURPosive	-nkV	-nkV
PRECAUTionary	-nyji	-nyji
SUCCessive	-nyjiku	-nyjiku
unmarked (UM)/SUBordinate	-nyV	-nyV

Table 3.16 : Paradigm of verbal inflections

Table 3.17: Inflected verbs
-----------------------------	--

	'hit'	'sit'	'break (itr)'
NONPAST	kunil	bunday	kadabal
PAST	kuniny	bundany	kadabany
IMP	kuni	bunday	kadabaka
IRR	kuninyaku	bundanyaku	kadabanyaku
PURP	kuninka	bundanka	kadabanka
PRECAUT	kuninyji	bundanyji	kadabanyji
SUCC	kuninyjiku	bundanyjik	kadabanyjiku
UM/SUB	kuninya	bundanya	kadabanya

3.8. 4.2 Functions of verbal inflections

The eight functions of the inflections listed above are as follows:

(i) NONPAST -I, -y

This inflection, which is identical to the conjugation marker, refers to an action or state in the present or future. To make reference to future explicit, a time word such as juma 'soonllater', wunkuny 'next day', etc. is added.

(144)	Warru-ngku	baya	nyanda-l.
	yg.man-ERG:pt(A)	firewood.ABS(O)	split-NONPAST
	'The young man is splitting firewood.'

(1 45) Jarramali juma kada-y. thunderstorm.ABS(S) soon come-NONPAST 'A thunderstorm will come soon.'

(ii) PAST -ny

A verb with past tense inflection may have perfective or imperfective implication. It can usually be inferred from the context whether an action or state that began in the past has been completed or is still going on at the time of the utterance.

(146)	Dingkar-angka	naybu	nyurrba-ny.
	man-ERG:pt(A)	knife.ABS(O)	grind/sharpen-PAST
	'The man sharpened the knife.'
(147)	Bama	nyunba-ny	kada-ny-baja
	Aborigine.ABS(S)	turn.back-PAST	come-PAST-again
	nyungu-wun-bu	bayan-ba.
	3sg-POSS-LOC	camp-LOC
	'The Aborigine turned back and came back to his camp.'

(1 48) Karrkay bambay-ma-ny. child.ABS(S) sick -INCHO-PAST 'The child fell ill.'

(iii) Imperative -¢ - -y - -ka

This form is typically used in an order or request addressed to a second person or persons. However, it has also been attested with first person dual or plural forms expressing an invitation or suggestion 'let's do ... !' (see §4.7 for imperative clauses).

The three allomorphs of imperative inflection are conditioned by conjugation class and the number of root or stem syllables of a verb.

Disyllabic roots in I-conjugation have zero inflection for imperative:

• ( 149) Yarraman kari kanga! horse.ABs(o) NEG chase.IMP 'Don't chase the horse!'
• (1 50) Jarruja dida! trousers.ABS(O) put.on.IMP 'Put your pants on!'

Three I-conjugation verbs have suppletive roots in imperative form:

Citation	Imperative
daji-l	daya	'give'
nyaji-l	nyaka	'see/hear'
mani-l	mana	'get'

These forms have been analysed in §3.8.3.2 (i) above.

Disyllabic roots in y-conjugation take imperative inflection -y, for example:

• (151) Dunga-y-baja! go-IMP-again 'Go back!'
• ( 1 52) Kari yirrka-y! NEG shout-IMp 'Don't shout!'

Since y-conjugation verbs have the same inflectional suffix for non-past tense and imperative, and the subject, i.e. the addressee, is often included in an imperative clause (see §4.7), the only certain distinguishing factor between a nonpast declarative and an imperative clause is the intonation contour (see §2.6.3).

Roots and stems with more than two syllables of either conjugation take the allomorph -ka, for example:

• (1 53) N gawa kararr-ba yijarri-ka! baby.ABS(O) blanket-LOC put-IMP 'Put the baby on the blanket!'
• (1 54) Mara julurri-ji-ka! hand.ABS(O) wash-REFL-IMP 'Wash your hands !'
• (1 55) Kuku-ku binal-ma-ka! language-DAT know-INCHO-IMP 'Learn language!'

Note that compounds and causative stems formed with -bunga-l, -man i-I, etc. (see §3.8.5.1 and §3.8.5.2) do not take suffix -ka but zero inflection for imperative. This means that only the number of syllables of the second (stem-forming) component is relevant, not the number of syllables of the whole stem.

( 1 56) Wumbul-bunga! hot -CA US.lMP 'Heat [it] up!'

Furthermore, causative stems formed with mani-l show the suppletive root mana ill imperatives, for example:

( 1 57) Kari daray-mana! NEG fall-CAUS.lMP 'Don't drop [it]!'

For these reasons, and because of the way in which they reduplicate, causative stems are most appropriately regarded as compounds rather than derivations (see §3.8.5. 1 ).

(iv) 1rrealis -nyaku

An irrealis verb form indicates that an action that should or could have taken place does or did not occur for some reason. 1rrealis verb forms are used in independent clauses, which may or may not be co-ordinated with another clause stating why the action is/was prevented. The irrealis inflection itself does not contain temporal reference; this must be inferred from context.

(158)	Ngayu	badur-iji	dunga-nyaku.
	1sg.NOM(S)	fishing.line-COMIT	go-IRR
			'I would go/would have gone fishing [but I cannot for some reason].'

(1 59) N gayu nuka-nyaku minya, kari yinya minya kabu. 1 sg.NOM(A) eat-IRR meat.ABS(O) but that.ABs(s) meat.ABS(S) rotten 'I would have eaten that meat, but it was rotten.'

(v) Purposive -nkV

The purposive inflection follows Kuku Yalanji vowel-harmony rules and has allomorphs -nka after final root or stem vowels a and i, and -nku after final vowel u. The Kuku Yalanji purposive inflection is presumably cognate with the widespread purposive inflection -gu in Australian languages.

A purposive verb form expresses 'desire or intention to Verb'. It may be used in independent or subordinate clauses and occurs obligatorily with adjectival particles (see §3.9.1 ).

• (1 60) N gayu mayi waju-nku. 1 sg.NOM(A) food.ABS(O) cook-PURP 'I want to cook some food.'
• (161) Nyulu binal dama-nka. 3sg.NOM(S) know spear-PURP 'He knows how to spear. '

( 1 62) Jana makarr wundi-ny kuyu wubul 3pLNOM(A) net.ABS(O) take-PAST fish.ABS(O) many.ABS(O) bangka-nka. catch/gather large quantities-PURP 'They took a net to catch a lot of fish.'

For a full discussion of purposive constructions see §4.4. 1.

(vi) Precautionary -nyji

The precautionary verb form indicates an undesirable event that should and can be avoided or prevented. A precautionary clause is typically subordinated to a main clause which describes the action that is or should be taken against the undesirable outcome.17

¹⁷ Dixon ( 1 980:380) uses the term "apprehensional inflection" and ( 1977:350) '''lest' subordinate clause" for Yidiny, which has the cognate inflections -ndi - -ldi - di. I prefer the term 'precautionary' for Kuku Yalanji because this construction normally states explicitly what precaution has to be taken against something.

• (1 63) N gayu jawun-unji dunga-y I sg.NOM(S) friend-COMIT:pt go-NONPAST 'I go with a friend lest I get lonely. ' milkabujar-ma-nyji. lonely-INCHO-PRECAUT
• ( 1 64) Jujubala waju jarramali kada-nyji! ironwood.ABS(O) burn.lMP thunderstorm.ABS(S) come-PRECAUT 'Bum some ironwood to prevent the thunderstorm from coming!'

Precautionary constructions are discussed in §4.4.2.

(vii) Successive -nyjiku

A successive verb form in a subordinate clause refers to the latter of two successive events, which have no causal connection. The preceding event is expressed in the main clause, for example:

• ( 1 65) Jana maYl kunba-y-mani-ny nganjan 3pl.NOM(A) food.ABS(O) finish-y-CAUS-PAST father.ABS(S) 'They had finished the food before [their] father woke up. ' wandi-nyjiku. wake.up-succ
• ( 1 66) Nyulu yirrka-ny wula-nyjiku. 3sg.NOM(S) cry. out-PAST die-SUCC 'He cried out before he died.'

For a discussion of successive constructions see §4.4.3.

The successive inflection -nyjiku and precautionary -nyji are similar in form. They also share the semantic component of 'following event', successive as an actual time reference and precautionary as an expected undesirable event. The cognate 'anticipatory' inflection -yiku in Guugu Yimidhirr can apparently imply either of these aspects (Haviland 1 979a:93).

(viii) Unmarked/subordinate inflection -ny V

Like purposive suffix -nkV this suffix follows vowel-harmony rules and has the allomorphs -nya after final root vowels a and i, and -nyu after root vowel u. This suffix is used in various functions, all of which have the common aspect of 'general subordination'.

In the Nyungkul dialect verbal suffix -nyV is used as an inflection that is apparently unmarked for tense or aspect to offset non-events or background information in narratives, i.e. 'general subordination' at discourse level. This discourse function is discussed in §5.5.1 .2. The Yalanji dialect does not make this distinction in discourse, but uses normal tense inflections. In both Yalanji and Nyungkul verbal suffix -nyV occurs in the following subordinate clauses: causal, simultaneous action, prior event, hypothetical reason.

In the causal subordinate clause -nyV is clearly an inflection since the verb + -nyV form may not be inflected further in any other way, for example:

( 1 67) N gayu bambay-ma-ny mayngku kayal nuka-nya. I sg.NOM(S) sick-INCHO-PAST mango.ABS(O) unripe.ABS(O) eat-sUB 'I became sick from eating unripe mangoes.'

A noun phrase in corresponding causal function requires ablative inflection. (See §4.1 .4.4 (iv) (d).)

In the other subordinate clauses, however, the verb + -nyV form is inflected for case. For this reason -ny V has to be regarded as a nominaliser in these instances as had been indicated in §3.2.3.6. Indeed, verbal inflection -nyV and nominaliser -nyV may be unrelated homophonous suffixes, but the basic function of 'subordination' common to all uses of -nyV seems to indicate that there is some connection. 18 Because of this common function all occurrences of -nyV in subordinate clauses are glossed as SUB (ordination), but in its nonevent marking capacity in Nyungkul narratives suffix -nyV is glossed as UM (unmarked inflection) to distinguish its use from subordinate clauses. Syntax and function of subordinate clauses with verb + -nyV forms are discussed in §4.4.4. 1-4.4.4.4.

3.8.5 Verbal stem formation

Kuku Yalanji has only relatively few verb roots, but several highly productive stemforming processes. Causative verb stems may be formed from any nominal, intransitive verb or adjectival particle by regular compounding with a causative root (see §3 .8.5. 1 ). Numerous other compounds are formed by combining nominals or particles with free-form verb roots, often resulting in idiomatic expressions (see §3.8.5.2). Inchoative verb stems may be derived from almost all non-verbal parts of speech with the stem-forming suffix -ma-l (see §3.8.5.3).

Apart from these processes which actually create new verbs, there are two syntactic (reciprocal and reflexive/general intransitive) and two non-syntactic derivations (reduplication and plural subject agreement), which are discussed in §3.8.5.4-3 .8.5.5.

3.8.5. 1 Causative compounds

Causative verbs in Kuku Yalanji can be classed into two categories according to the result that they imply. One type of causative compound refers to the causation of a state, the other to the causation of an action referred to by an intransitive verb. The former consist of a nominal root or stem, or an adjectival particle plus 'state causative' component -bunga-l or -kanga-l; the latter of an intransitive verb root plus 'action causative' component -mani-l. (The causation of a transitive action cannot be expressed by a verbal compound, but requires an explicit statement as to how the action was caused, e.g. 'he told him to do something' or 'he forced him to do something', etc.)

Causative components -kanga-l and -mani-l also occur as free-form roots meaning 'chase, disturb' and 'get' respectively. Suffix -bunga-l, on the other hand, cannot be used independently but the morphological behaviour of -bungal forms in reduplication and imperative is exactly like that of the other causative compounds or of idiomatic compounds. (For imperative forms see §3.8.4.2. (iii), for reduplication §3.8.5.5. (ii).) It would therefore be reasonable to regard -bunga-l as a compound verb-forming root with the semantic content 'make into Noun/make Adjective' rather than a derivational suffix.

¹⁸ Consider also the similarity between -nyV and irrealis -nyaku, even though the latter apparently does not follow vowel harmony rules. But the two forms could well be related (cf. irrealis and subordinate function of one prefix form in Mangarayi (Merian 1 98 1 ).

R. Hershberger's (1 964b:37) interpretation of -nyV as a 'verb serialiser' could correspond to the discourse function of this suffix in Nyngkul, although I do not know from which context her example was taken.

98 Chapter 3

(i) State-causative compounds with -bunga-l or -kanga-l

State-causative verbaliser -bunga-l has a reduced allomorph -ba-l, sometimes also -buwa-I, in rapid speech, but only the full form is used in reduplication. The form -bunga-l can attach to a nominal root or stem or adjectival particle.

Simple root + -bunga-l:

dandi	'hard'	dandi-bunga-l	'make hard'
yarkin	'corpse'	yarkin-bunga-l	'kill'
manyarr	'wife'	manyarr-bunga-l	'cause to be a wife (i.e. marry)'
binal	'know'	binal-bunga-l	'teach'

Reduplicated root + -bunga-l:

kaka-kaka	'painful'	kaka-kaka-bunga-l	'make painful'
bajay-bajay	'exhausted'	bajay-bajay-bunga-l	'exhaust someone'

Reduplication of the actual compound, which involves only the second component, is a different matter. While the above reduplicated nominal roots refer to the state that is caused, reduplication of the second element of the compound implies prolonged or repeated causation of a state, for example:

(168)	Jana	kuku	muwul-bunga-l-bunga-l.
	3pl.NOM(A)	word.ABS(O)	secret-CAUS-l-REDUP-NONPAST
	'They keep the word/message secret.'

A causative compound with a reduplicated stem may be further reduplicated, e.g. kakakaka-bunga-l-bunga-l 'keep making painful'.

Compound + bunga-l:

jiba-wulawula	'sweet'
jiba-wulawula-bunga-l	'sweeten'

Comitative stem + bunga-l:

ngurrka-ji	'have a scar'	ngurrka-ji-bunga-l	'scar someone'
bubu-ji	'soiled, dirty'	bubu-ji-bunga-l	'make dirty'

The choice between -bunga-l and -kanga-l for a state-causative compound is made on semantic grounds. The form -bunga-l is the unmarked state-causative verbaliser, while -kanga-l, which also functions as an independent transitive verb meaning 'chase, disturb', implies in a compound verb 'causation of physical or emotional commotion', and can occur only with simple nominal roots.

ngirray	'weep'	ngirray-kanga-l	'evoke weeping'
yinil	'afraid, fright'	yinil-kanga-l	'evoke fright, frighten'
bunjay	'laughter'	bunjay-kanga-l	'evoke laughter'
burkur	'dirty, muddy (water)	burkur-kanga-l	'make water dirty by stirring it up'

Because of the semantic distinction between -bunga-l and -kanga-l some compounds can involve only one or the other, for example:

'evoke weeping'	ngirray-kanga-l	NOT	ngirray-bunga-l
'induce bleeding'	mula-kanga-l	NOT	mula-bunga-l
'evoke anger'	kuli-kanga-l	NOT	kuli-bunga-l
'make dopey'	juwa-bunga-l	NOT	juwa-kanga-l
'make slow'	waymbul-bunga-l	NOT	waymbul-kanga-l

'teach' binal-bunga-l NOT binal-kanga-l

Other compounds, on the other hand, may take either -bunga-l or -kanga-l as second component according to context, for example:

• ( 1 69) a. Karrkay-ngka kambi mumbu-bunga-ny. child-ERG:pt(A) clothes.ABS(O) dirty-CAUS-PAST 'The child made the clothes dirty.'
  • b. Karrkay-ngka bana mumbu-kanga-ny. child-ERG:pt(A) water.ABS(O) dirty-CAUS-PAST 'The child made the water dirty.' (i.e. by disturbing it)
• ( 1 70) a. Maral-angka mayi jirra-bunga-ny. girl-ERG:pt(A) food.ABS(O) overdone-CAUS-PAST 'The girl overcooked the food.'
  • b. Baya jirra-kanga-¢-baja! fire.ABS(O) overdone-CAUS-IMP-again 'Stoke the fire again !'
  • c. Yundu nganya kari jirra-kanga! 2sg.NOM(A) I sg.ACC(O) NEG overdone-CAUS.lMP 'Leave me alone!' (i.e. don't stir up my temper)

(ii) Action-causative compounds with -mani-l

Root -mani-l, appended to an intransitive verb, creates a transitive causative verbal compound. In the formation of such a compound a link morpheme -y- is inserted between the two roots after final vowels a or u; .this link morpheme is the same for intransitive roots of either conjugation. Thus the formula for an action causative compound verb is:

intransitive verb -at -u# +y +mani-l

Examples of verb root + mani-l are:

BUT

dara-y	'fall'	dara-y-mani-l	'cause to fall, drop'
jana-y	'stand'	jana-y-mani-l	'cause to stand'
wala-l	'enter'	wala-y-mani-l	'cause to enter'

(171) Waybala-ngka maral school-bu white.man-ERG:pt(A) girl.ABS(O) school-LOC 'The white man made the girl go to school.' wala-y-mani-ny. enter-y-CAUS-PAST Examples of compound verb + mani-l are:

j ulbarrwarri-y	'slip'	julbarr-warri-mani-l	'cause to slip'
walu-dunga-y	'be annoyed'	walu-dunga-y-mani-l	'cause to be annoyed'

I presume that the action-causative verbaliser -mani-l is related to the verb root mani-l 'get' because the imperative form of action causative compounds uses the suppletive form of mani-l 'get', as injana-y-mana 'make (it) stand up!'

In Guugu Yimidhirr the monosyllabic roots maa-naa 'get' and causative -ma-naa appear to be similarly related. Both take irregular past tense inflection -ni, e.g. maa-ni, and the same suffix for stem formation before further inflections (Haviland 1 979a:84-85).

One idiomatic use of action-causative -mani-l with a transitive verb must be noted: ngaki-l 'hide' + -mani-l means 'elope', for example:

( 1 72) Nyulu ngaki-mani-ny yinya jalbu. 3sg.NOM(A) hide-CAUS-PAST that.ABs(O) woman.ABS(O) 'He eloped with that woman.'

R. Hershberger ( 1 964b:50) states that if an action-causative verb stem is "changed back to intransitive" the causative verbaliser "is dropped and the subject changes to locative" (my emphasis). This observation is true only in so far as action-causative stems, like other transitive verbs, may be passivised, in which case the agent takes locative inflection. However, the causative root -mani-l is not deleted, but the verb stem intransitivised with suffix -ji-, for example:

( 1 73) Karrkay ngamu-ndu school-bu wala-y-mani-ji-ny. child.ABS(S) mother-LOC school-LOC enter-y-CAUS-ITR-PAST 'Mother mistakenly sent the child to school (it was still too young). '

(For passive formation and functions see §4.2.4.3.)

3.8.5.2 Idiomatic verbal compounds

Apart from the very productive compound formation with causative roots, many other compounds are formed by combining a nominal root with a verb. Such compounds are always inseparable and usually have an idiomatic meaning that is often only loosely connected with the semantic content of the two components.

Body-part terms and the nominal root wawu 'breath, spirit' are most often used in verbal compounds. (For use of body parts in nominal compounds see §3.2.3. 1 0.) Other nouns occur less frequently and adjectives least of all. To give a few examples:

body part + verb:

mara-warri-y	(hand-run-y)	'crawl'
jiba-warri-y	(liver-run-y)	'get startled, get a fright'
milka-wula-y	(ear-die-y)	'forget'
bujil-janji-l	(nose-bathe-l)	'drown (itr)'
walu-dunga-y	(face-go-y)	'faint'
dukul-warri-y	(head-run-y)	'throw a tantrum'

wawu 'breath, spirit' + verb:
wawu-wula-y	(-die-y)	'be exhausted, thirsty'
wawu-yimba-l	(-go straight through-I)	'cross-examine'
wawu-yilba-l	(-throw-I)	'sigh'
other nouns + verb:
burri-duda-l	(song-beat-I)	'beat time to corroboree'
jila-warri-y	(sweat-run-y)	'perspire'
ngirray-warri-y	(weeping(N)-run-y)	'burst into tears'
kandarr-warri-y	(burp(N)-run-y)	'to burp'
buya-warri-y	(fart(N)-run-y)	'to break wind'
nyunga-dama-l	(thought-spear-l)	'imagine something'
kunji-dama-l	(sneeze(N)-spear-l)	'to sneeze'
adjective + verb:
burra-warri-y	(stale-run-y)	'stagger, lurch'
julbarr-warri-y	(slippery-run-y)	'slip'
buyun-dama-l	(bad-spear-l)	'spoil'

As can be seen in the above examples the verb root warri-y 'run' is frequently used in the formation of compounds. Dama-l 'to spear' is also used relatively often, although its semantic content in compounds is somewhat opaque, except in kunji-dama-l 'to sneeze' where it may refer to the motion performed in sneezing, which is similar to the motion of the arm when throwing a spear. Dama-l also occurs in the only attested verb + verb compound walngka-n-dama-l (hang-spear) 'hang up to dry' where it may have similar reference to the motion involved in throwing, say, clothes, over a branch or line.

3.8.5.3 Inchoative stems

Inchoative verb stems may be formed from nominals with the derivational suffix -ma-l or with the intransitive form -maniji-y.

(i) Inchoative stems with -ma-l

Derivational suffix -ma-l is a highly productive verbaliser which can combine with virtually any nominal root, adjectival particle and, where semantically plausible, with time and location words, for example:

yalbay	'big'	yalbay-ma-l	'grow'
dandi	'hard'	dandi-ma-l	'become hard'
kulji	'stone'	kulji-ma-l	'turn into stone'
kalka	'spear'	kalka-ma-l	'turn into a spear'
manyarr	'wife'	manyarr-ma-l	'become a wife, marry'
binal	'know'	binal-ma-l	'learn'
yilay-yilay	'late afternoon'	yilay-yilay-ma-l	'become late afternoon'
kuwa	'west'	kuwa-ma-l	'move westward'

• ( 1 74) Wungar ngalangala-ma-ny. sun.ABS(S) faint.red-INCHO-PAST 'There was an eclipse of the sun.' (lit: The sun became faint red.)
• (1 7 5) Juku ngunjil-ma-ny. tree/wood.ABS(S) charcoal-INCHO-PAST 'The wood turned into charcoal. '
• ( 1 76) Nyulu dunga-ny-baja naka-ma-ny. 3sg.NOM(S) gO-PAST-again east-INCHO-PAST 'He went on again and moved eastwards. '

Inchoative stems with -ma-l are clearly derivations and not compounds because they take imperative inflection -ka, used with stems of more than two syllables, and reduplicate as a whole.

Cognates of Kuku Yalanji -ma-l can be found in all its related languages: -ma-l in Guugu Yimidhirr (Haviland 1 979a:65), -mayi-y in Djabugay (Hale 1 976b:239; Patz 1991:272) and -maji-n, with only very limited function, in Yidiny (Dixon 1977a:368).

Nyungkul speakers may use the inchoative verbaliser -ma-l with inflected nouns, for example:

( 1 77) Nyulu wungar-a-ma-l. 3sg.NOM(S) sun-LOC-INCHO-NONPAST 'He is moving into the sun. '

but Yalanji speakers do not favour such constructions (see also ( 1 80) and following comment).

(ii) Inchoative compounds with -maniji-y

Unlike inchoative stems with -ma-l, inchoative forms with -maniji-y must be regarded as compounds because their morphological behaviour in imperative and reduplication corresponds to that of other compounds. The form -maniji-y could be an intransitivised form (see §3.8.5.4(ii)) of causative verbaliser -mani-l, but note that -mani-l, unlike -maniji-y, does not occur with nominals.

The form -maniji-y may attach to inflected nouns or comitative stems as well as roots, but its use is much more limited than that of the derivational suffix -ma-l. It can imply 'become ^Xowing to external circumstances' but this interpretation does not always apply. Some examples are:

• ( 1 78) N gayu wulbuman-maniji-ny. I sg.NOM(S) old.woman-INCHO-PAST 'I became an old woman.'
• (1 79) Juma Leonie dunyu-nji-maniji-y. later Leonie.ABS(S) husband-COMIT-INCHO-NONPAST 'Eventually Leonie will get married.'
• ( 1 80) Ngayku bayan wungar-a-maniji-y. I sg.POSS.ABS(S) house.ABS(S) sun-LOC-INCHO-NONPAST 'My house will move into the sun. '

Note that the last example is not a passive clause with 'sun' as agent in locative case because wungaramanijiy is inseparable.

Such inchoative forms are used only infrequently and inchoative derivations with -ma-l are generally preferred. Possibly, an original semantic distinction between the two inchoative forms is becoming blurred, with -ma-l taking over as general inchoative stem-forming suffix.

3.8.5.4 Syntactic derivations

There are two derivational processes in Kuku Yalanji which change a transitive verb into an intransitive one. One of these is 'reciprocal', the other is 'reflexive' which is also used to intransitivise verbs in other than reflexive contexts. The change in transitivity goes together with a change in conjugation class: an intransitive stem derived from a transitive l-conjugation verb always belongs to the y-conjugation.

(i) RECIProcal -wa-

A reciprocal verb stem is used where two or more participants are involved in the same action on each other. In such constructions the respective agent(s) and transitive object(s) are not referred to individually in these functions, but conjoined in one noun phrase which is in intransitive subject function, and the verb must be intransitivised with 'reciprocal' stemforming suffix -wa-. (For more detail about such constructions see §4.2.3.) Some examples of reciprocal stems are:

Transitive verb		Reciprocal stem
nyaji-l	'see'	nyaji-wa-y	'see each other'
kunja-l	'call, summon'	kunja-wa-y	'call each other'
binal-bunga-l	'teach'	binal-bunga-wa-y	'teach each other'
murru-kanga-l	'scold'	murru-kanga-wa-y	'scold each other'

(1 81) Jarba bula kaja-wa-y. snake.ABS(S) 3du.NOM(S) tie-RECIP-NONPASf 'The two snakes wrap around each other. ' (i.e. mate)

(182)	Jana	bama	jalbu-jalbu	kuni-wa-y
	3pl.NOM(S)	person.ABS(S)	woman-REDUP.ABS(S)	hit-RECIP-NONPAST
	jana	kuli-kanga-wa-y	jurbu-bu.
	3pl.NOM(S)	wild-CAUS-RECIP-NONPAST dance-INST
	'The women are fighting; they stir each other up with a dance.'

Reciprocal stems may of course be reduplicated to indicate continuous action, e.g. kuji-nkuji-wa-y 'keep waiting for each other/looking after each other'. But since reciprocal stems inherently imply a repeated action, reduplication is only used where an abnormally long continuation needs to be indicated. (For reduplication of reciprocal stems see §3.8.S.S (ii).)

Two reciprocal stems, kuni-wa-y 'hit each other' and balka-wa-y 'talk to each other' can occur outside the context of explicit reciprocity, meaning 'fight' and 'chat' respectively, for example:

1 04 Chapter 3

( 1 83) Nyulu wawu kari kuni-wa-nka. 3sg.NOM(S) want NEG hit-RECIP-PURP 'He does not want to fight.' (184) Dingkar yinya balka-n-balka-wa-y man.ABS(S) that.ABS(S) speak-n-REDUP-RECIP-NONPAST nyulu kuku-baka. 3sg.NOM(S) language-excess 'That man talks a lot, he is a chatterbox.'

This would mean that although these forms are historically reciprocal derivations, and can be used as such as in (1 82), they also represent separate lexemes with an extended semantic content.

In the Yalanji dialect it is possible to use the reciprocal suffix with some intransitive l-conjugation roots apparently in free variation with the plural subject marker -ri- (see §3.8.5.5 (i» . There is no apparent semantic motivation for this, nor are there syntactic implications.

(ii) REFlexive/general intransitive (ITR) suffix -ji-

The primary use of stem-forming suffix -ji- is in reflexive constructions. If an agent performs an action on him/herself the coreferential transitive object must be omitted in Kuku Yalanji. The verb is intransitivised with reflexive suffix -ji- and the agent appears in intransitive subject function. (For more syntactic details see §4.2.4.2.) Some examples of reflexive stems are:

Transitive verb		Reflexive stem
yaka-l	'cut'	yaka-ji-y	'cut oneself'
jubi-l	'squeeze, pinch'	jubi-ji-y	'pinch oneself'
kuli-kanga-l	'excite'	kuli-kanga-ji-y	'excite oneself'
jana-y-mani-l	'make stand up'	jana-y-mani-ji-y	'make onself stand up'
(185) Dingkar julngka-ji-y warrma-ka. man.ABS(S) paint-REF-NONPAST corroboree-DAT 'The man paints himself for the corroboree.'

Similar to fixed reciprocal stems there are a number of reflexive stems which have undergone a semantic shift from the meaning of the corresponding root and are used as separate lexemes in their own right. As I was able to infer from informants' reactions, some of these reflexive stems are hardly connected to the semantic content of the root in the speaker's mind. Thus it is acceptable to say:

( 1 86) Buji jika-ji-y. cat.ABS(S) stretch-REF-NONPAST 'The cat is stretching itself.'

but quite impossible to say:

( 1 87) *Ngayu buji 1sg.NOM(A) cat.ABS(O) stretch-NONPAST 'I stretch the cat. ' jika-l.

since jika-l refers exclusively to the action of straightening a spear by heating it in a fire and stretching, bending and turning it. A list of attested reflexive stems that can be regarded as separate lexemes, together with their corresponding roots, is given below:

Root		Fixed reflexive stem
balka-l	'make'	balkaji-y	'get born'
jika-l	'straighten a spear'	jikaji-y	'stretch oneself'
jurrki-l	'move something'	jurrkiji-y	'move house/camp (oneself)
murni-l	'stir'	murniji-y	'spin around/take a detour'
nubi-l	'search for'	nubiji-y	'be lost'

Stem-forming suffix -ji- is also used to intransitivise verbs in passive and antipassive constructions in Kuku Yalanji. Syntax and functions of such constructions are considered in §4.2.4.3 and §4.2.4.4. (As a general intransitivising suffix in passive and antipassive -ji- is glossed as ITR.)

Kuku Yalanji -ji- is presumably a reflex of a pA verbal suffix *DHirri-y and as such a variation of probably the most widespread and productive verbal suffix to be found in eastern and central languages of Australia (see Dixon 1 980:447-449). All reflexes of *DHirri-y, e.g. -Vdhi in Guugu Yimidhirr, -yi- in Djabugay, Kalkatungu and Lardil, Vji-n in Yidiny and -yirri-y in Dyirbal, have a basic intransitivising function, although the semantic implications differ somewhat from language to language.

3.8.5.5 Non-syntactic derivations

Reduplication and plural subject agreement are non-syntactic derivations which, ^m contrast to syntactic derivations, have no impact on the transitivity class of the verb.

(i) Plural subject marker -ri-

This suffix is used on intransitive roots or stems of either conjugation to indicate verb agreement with a plural intransitive subject. A -ri-stem cannot, however, indicate by itself plurality of the subject; that is, the subject must be marked for plural by reduplication or a plural pronoun. (For plural formation see §3.2.3.5). The use of -ri- for plural agreement is optional; it appears to be a redundant form.

Root or stem		Plural subject form
dunga-y	'go'	dunga-ri-y
janji-l	'bathe'	janji-ri-y
waju-ji-y	'burn (REF)'	waju-ji-ri-y
kuni-wa-y	'hit (RECIP)'	kuni-wa-ri-y

Examples of sentences with plural subject/verb agreement are:

( 1 88) Jana bama jabajaba wandi-ri-ny. 3pl.NOM(S) Aborigine.ABS(S) all.kinds.ABS(S) emerge-PLS-PAST 'All sorts of people turned up. '

• (1 89) Yurra kada-ri-ka! 2pl.NOM(S) come-PLS-IMP 'You lot, come here!'
• (190) Karrkay-karrkay bayan-ba ngaki-ji-ri-y. child-REDUP.ABS(S) house-LOC hide-REF-PLS-NONPAST 'The children hide in the house. '

gaki-ji-ri-y. hide-REF-PLS-NONPAST

There are no corresponding verb devices to indicate plural transitive subject or intransitive object. Here, plurality is expressed only by the noun or pronoun form.

(ii) Reduplication

Verbal reduplication serves to indicate an ongoing, repeated or habitual action and/or a certain intensity in action, e.g. the reduplicated form of 'see' can imply 'look thoroughly, examine' or just 'keep looking'. In order to express excessively long duration, a verb is simply repeated a number of times without using reduplicated stems; the last vowel of the last verb is usually lengthened considerably, for example:

(191) Jana jalbu-ngku duda-l duda-l duda-l dudaaa-l. 3pl.NOM(A) woman-ERG:pt(A) pound-NONPAST ... 'The women pound and pound and pound and pound.' (i.e. preparing zamia nut)

With verbs of rest 'sit', 'lie' and 'stand' the reduplicated form refers to the state of 'sitting' (also 'living at'), 'lying' (also 'sleeping') and 'standing', whereas the simple root implies 'sit down, lie down, stand up' (i.e. get into posture).

There are different methods of reduplication for simple roots and inchoative stems, other derived stems, and compounds.

(a) Reduplication of simple roots and inchoative stems

The regular reduplication process of these involves repetition of the whole root or stem joined by a f ormal link morpheme:

root/stem + link + root/stem + inflection

The link morpheme is -l- for l-conjugation verbs and -n- for y-conjugation verbs.

Reduplicated l-conjugation verbs:

dinda-l-dinda-l	'keep roasting'
karrba-l-karrba-l	'keep holding on'
wukurri-l-wukurri-l	'keep following'
kima-ma-l-kima-ma-l	'keep getting soft'

Inchoative stems are generally reduplicated as a whole as in the above example, but many speakers, particularly younger ones, reduplicate such forms analogous to causative or other compounds; that is, they repeat only the stem-forming suffix as in kima-ma-l-ma-l (see R. Hershberger 1 964b:43). Most of the older informants regard the latter form as incorrect, even if they themselves use it frequently.

The three verbs that have suppletive roots for use with imperative inflection also require these alternative roots in a reduplicated imperative form, but not with other inflections:

nyaji-l-nyaji-l	'keep looking'
daji-l-daji-l	'keep giving'
mani-l-mani-l	'keep taking'

nyaka-l-nyaka daya-I-daya mana-I-mana

'keep looking!' 'keep giving!' 'keep taking!'

Examples of reduplicated y-conjugation verbs:

dunga-n-dunga-y	'keep going'
yirrka-n-yirrka-y	'keep shouting'

The linking -l- in reduplicated l-conjugation verbs could represent the conjugation marker, but linking -n- in y-conjugation verbs cannot be explained synchronically. But note that the conjugation marker of the predominantly intransitive verb class in Yidiny is - $n$ (Dixon 1977:207). Since y-conjugation verbs in Kuku Yalanji are all intransitive, this - $n$ - may be a remnant of a former n-conjugation which does not now exist in Kuku Yalanji.

Six y-conjugation roots show only partial reduplication of either the first or the second syllable. These are:

(i) Reduplication of first syllable

badi-y	'cry, weep'	banbadi-y	'keep crying, weeping'
jana-y	'stand (up)'	janjana-y	'keep standing'
kada-y	'come'	kankada-y	'keep coming'
warri-y	'run'	wanarri-y	'keep running'

Some speakers make a distinction between the partially reduplicated form wanarri-y 'perpetual motion, e.g. of running water', and a fully reduplicated form warri-n-warri-y 'continuous running by human or animal'.

(ii) Reduplication of second syllable

bunda-y	'sit (down)'	bundanda-y	'keep sitting, live at'
wuna-y	'lie (down)'	wunana-y	'keep lying, sleep'

Syllable reduction in partial reduplication has been discussed in §2.5.3.

(b) Reduplication of other than inchoative stems

In reduplication of reciprocal, reflexive/general intransitive and plural subject verb stems it is only the root that is reduplicated. The derivational suffixes -wa-, -ji- and -ri- are then appended to the reduplicated stem. However, in reciprocal and reflexive/general intransitive stems, which are formed from transitive l-conjugation roots, the link component is not -l-, but - $n$ - according to the conjugation class of the derived stem, i.e. y-conjugation. The same applies for intransitive l-conjugation roots (see §3.8.3.1):

Root		Reduplicated reciprocal stem
kuni-l	'hit'	kuni-n-kuni-wa-y	'keep hitting each other'
bayka-l	'bite'	bayka-n-bayka-wa-y	'keep biting each other'
daka-l	'climb (itr)'	daka-n-daka-wa-y	'keep climbing together'

Root		Reduplicated reflexive/intransitive stem
baka-l	'poke'	baka-n-baka-ji-y	'keep poking oneself'
yaka-l	'cut'	yaka-n-yaka-ji-y	'keep cutting oneself'
nuka-l	'eat'	nuka-nuka-ji-y19	'keep eating (itr)'
Root		Reduplicated plural subject stem
dunga-y	'go'	dunga-n-dunga-ri-y	'keep going'
wuri-y	'dance'	wuri-n-wuri-ri-y	'keep dancing'

(c) Reduplication of compounds

In compound verbs it is only the second component which is reduplicated, for example:

Compound	Reduplicated form
wumbul-bunga-l	wumbul-bunga-l-bunga-l
dara-y-mani-l	dara-y-mani-l-mani-l
ngirray-kanga-l	ngirray-kanga-l-kanga-l
buyun-dama-l	buyun-dama-l-dama-l

3.B.6 Adverbial forms

A verb may be modified by an adverbial particle or an adverb-like form that is derived from an adjective. None of these forms have inflectional possibilities.

3.8. 6.1 Adverbial particles

Kuku Yalanji has a set of particles that may accompany a verb and in some way describe the action referred to by the verb. These particles are listed below, grouped into semantic fields:

Posture (rest)
jangka	'upright, straight'
jarrbarr	'prone, on stomach'
mingki	'crouching, squatting'
Direction (motion)
wadu(wadu)	'astray, off course'
maku	'straight towards'
walku	'past'

¹⁹ Since Kuku Yalanji does not allow geminate consonants (see §2.2.1) link - $n$ - and root-initial $n$ contract to single $n$ . The link component is therefore not represented in such forms.

Manner
ngambuy	'fittingly, appropriately'
nganjarr(nganjarr)	'discreetly, with a hidden purpose'
wudu	'absorbed in some activity'
ngura	'(too) fast'
yarngkay	'secretly'
nguymal	'thoroughly, truly'
Others
nguyarr	'not quite (i.e. an action occurs, but not totally)'
murruji	'cannot'
kanban	'by mistake'
nganga	'briefly'
jururr(jururr)	'repeatedly; with motion verbs: one after the other'

These forms may be reduplicated and/or take the manner suffix -ku like adjectives in adverbial function (see §3 .8.6.2). (Where a reduplicated form is given in the above list, this is the normally used form.) Nevertheless these particles are distinct from adjectives in that they (a) can occur in simple root form; (b) do not have to take manner suffix -ku; and, more importantly, (c) cannot be part of a noun phrase and take case inflections like adjectives. Adverbial particles may be placed anywhere in a clause but the preferred position is immediately before or after the verb.

The following sentences illustrate the use of some of these particles:

• ( 1 92) Nyulu kadar kuni-ny nguymal(-ku). 3sg.NOM(A) wallaby.ABS(O) hitlkill-PAST thoroughly 'He killed the wallaby with one shot.' (i.e. he made a thorough job of it)
• ( 1 93) Buliman kada-ny nganjarrnganjarr. police.ABS(S) come-PAST discreetly 'The police came discreetly.' (i.e. plain-clothes police)
• ( 1 94) Kari bunda-y! Ngarri buyun. Yunu-nji nguyarr dara-y. NEG sit-IMP leg.ABS(S) bad 2sg-COMIT not.quite fall-NONPAST 'Don't sit down! [The bench] has a rotten leg. It will tip over with you.' (i.e. not really fall, but just tip to one side)
• ( 1 95) Ngayu juku kulji-bu nguyarr kuni-ny. I sg.NOM(A) tree.ABS(O) stone-INST not.quite hit-PAST 'I did not quite hit the tree with the stone (but the stone may have touched it in passing).'
• ( 1 96) Yundu ngambuy bundanda-y bayan-ba. 2sg.NOM(S) appropriately sit.REDUP-NONPAST house-LOC 'You look just right living in that caravan.' (i.e. you fit in appropriately)

3.8. 62 Derived adverbs

Adverbs may be derived from adjectives or comitative or privative stems by the following processes:

• (a) attaching a 'manner' suffix -ku (to be glossed as ADVerb)
• (b) reduplication
• (c) reduplicated form + -ku.

Process (a) is the most productive, (b) is less often and (c) very rarely used. With comitative and privative stems reduplication is not favoured. Some examples are:

jinbal	'quick'	jinbal-ku	'quickly, fast'
		jinbal-jinbal(-ku)
buyun	'bad'	buyun-ku (etc.)	'badly'
mumbar	'neat, firm'	mumbar-ku	'neatly, firmly'
kuli-ji	'angry, wild'	kuli-ji-ku	'angrily'
kiru-kari	'stupid'	kiru-kari-ku	'stupidly'

• (1 97) Yinya karra mumbar-ku wunana-y. that.ABS(S) rope.ABS(S) neat-ADV lie.REDUP-NONPAST 'The rope is lying coiled up. ' (lit. neatly)
• (198) Yarraman mukul waymbul-ku dunga-y. horse.ABS(S) old.ABS(S) slow-ADV go-NONPAST 'The old horse is walking slowly.'

In transitive clauses, the derived adverb is based on an adjective + 'neutral' ergative/ instrumental form, for example:

(199)	Nyulu	ngarrbal	jinbal-da-ku	wukurri-ny.
	3sg.NOM(A)	stranger.ABS(O)	quick-ERG/INST-ADV	follow-PAST
	'He quickly followed the stranger.'

(200) Diburr mumbar-abu-ku bulin-ba yijarri-ka! egg/s.ABS(O) careful-ERG/INST-ADV plate-LOC put-IMP 'Put the eggs carefully on the plate!'

It is not clear whether the inflection represents ergative case agreement with the agent, or instrumental. The former is commonly found in Western Desert languages, such as Yankunytjatjara (Goddard, pers. comm.). The interpretation of ergative case agreement is supported by the fact that such inflected forms occur only in transitive sentences. (Note also that some other modifiers such as possessive forms, for instance, take always 'neutral' inflection, regardless of the type of inflection on the head noun.)

A commonly used adverb is based on the suppletive demonstrative root yala 'this' which also functions as a comparison particle 'like, same' (see §3.9.2).

(201) Yala-ku ba lka I like-ADV speak/make-IMP 'Speak like this/make [it] like this !'

Adverbial suffix -ku may be attached to inflected nouns to form adverbials of time, for example:

(202) N gayu warru-ku Palm Island bundanda-ny. I sg.NOM(S) yg.man.ABS-ADV Palm.lsland sit.REDUP-PAST 'As a young man I stayed at Palm Island.'

(203) N gayu karrkay-nyja-ku minya kadar I sg.NOM(A) child-ERG-ADV meat.ABS(O) wallaby.ABS(O) ngangkin nuka-l-nuka-ny. porcupine.ABS(O) eat-l-REDUP-PAST 'As a child I used to eat wallaby and porcupine.'

However, it is possible, that -ku in this context is not the adverb-forming suffix but the homophonous emphatic clitic -ku. The latter could be interpreted as highlighting the temporal contrast implied in these sentences, e.g. 'As a young man (which I am no longer) I stayed at Palm Island'. Compare R. Hershberger's ( 1 979:55) attested 'adverbial' suffix -ku with an inflected nominal form:

(204) Dingkar-angka nyunguniny yarkin-ka-ku kuni-ny. man-ERG:pt(A) 3sg.ACC(O) corpse-DAT-ADV(?) hitJkill-PAST "The man murdered him (intended to kill)." (translation from R. Hershberger)

Again, -ku could be interpreted as either adverbial suffix, 'He hit/injured him deadly', or as emphatic clitic, 'He hit him quite dead'.

3.9 Free-form particles

Apart from the adverbial particles discussed in §3.8.6.1 above, Kuku Yalanji employs a number of non-inflected free-form particles in various functions at clause and sentence level. We can distinguish four groups: adjectival, comparison, negation and discourse particles.

3.9. 1 Adjectival particles

There are four adjectival particles:

binal	'know (about) something'
juburr	'be good at something'
yinil	'be afraid'
wawu	'want, like'

All of these are used in predicator function with a dative noun or purposive verb as complement (see §4.9. 1 ). The reasons for classification as adjectival rather than verbal particles are as follows:

(a) All these particles may form causative and inchoative verb stems with -bunga-l and -ma-l (see §3.8.5.1 and §3.8.5.3), which are used to verbalise nominals:

yinil-bungal	'frighten someone'20
binal-mal	'learn [about] something'

Verb stems derived from wawu 'want, like' show a vestige of a comitative suffix (see §3.2.3.3) and have somewhat extended semantic content:

wawurrwawurr-mal	'be happy; decide'
wawurrwawurr-bungal	'make happy'

Wawu is also a noun meaning 'breath, spirit, soul'. Note that the cognate noun in Guugu Yimidhirr with the same meaning must take comitative to express 'want, like'.

• (b) All these particles, except for juburr, form privative stems (see §3.2.3.4), e.g. wawukari 'don't want, don't like'. (possibly the privative form of juburr 'good at' is not used because there is a lexical alternative: manjarr 'no good at, lazy'.)
• (c) These particles, except for wawu, may take the degree suffix -bajaku (see §3.2.3 .8) which places them in the adjective rather than in the noun category. But note that these particles do not take case inflections and can therefore not be regarded as 'proper' adjectives.

juburr-bajaku 'be very good at' binal-bajaku 'know very much, know exactly'

(To intensify wawu the quantifier jirray is used: wawu jirray 'like a lot, love'.)

3.9.2 Comparison, negation and discourse particles

These particles, in contrast to those discussed in the previous subsection, have no regular derivational possibilities. Only some isolated lexical items have one or the other of these particles as their basis. The functions of the particles listed below are described in following chapters as indicated.

(i) Comparison particle

Particle yala '(be) like, (be) the same' (glossed as COMP) may be related to the suppletive demonstrative root yala 'this'. The syntactic function of particle yala is discussed in §4.9.3. When used in reference to appearance this particle co-occurs with walu 'face' in either order, for example:

(205) Nyulu yala walulwalu yala bikibiki. 'He looks like a pig.'

Yala may also take the emphatic clitic -rrku (see §3. 1 0.2): yalarrku 'just the same; also', for example:

(206) Kanbal bubu yala-rrku. some.ABS(S) place.ABS(S) COMP-EMPH4 'Some places are just the same [as this place]. '

²⁰ Yinil 'be afraid' is used in free variation with a comitative form yinil-ji or yini-ji which, however, may not be verbalised.

(207) Wulman yala-rrku manyarr-anka badi-ny. old.man.ABs(S) COMP-EMPH4 wife-DAT:pt cry-PAST 'The old man also cried for his wife [just like the others].'

(ii) Negation particles

These particles include kari 'not' (NEG) and balu 'desist, give up' (NEG2)' Particle kari may take clitics -ku 'general emphasis' and -rrku 'totality' (see §3.10.2). Kariku has emphatic or temporal meaning according to context, that is 'not all' or 'not yet'. Karirrku implies 'not really, not altogether', for example:

(208) N gayu kari-rrku warngku-wuna-ny. 1 sg.NOM(S) NEG-EMPH4 sleep-lie-PAST 'I did not really sleep (I just dozed).'

(The functions of kari as negation particle are discussed in §4.9.3.)

Balu may take clitic -da, actually a temporal clitic (see §3.10.1), but baluda is used in the sense of 'let it be, don't do what you are about to do'. The use of balu as negation particle is discussed in §4.9.2.

(iii) Discourse particles

Kari and balu may be used not only in reference to part of a clause, for instance the predicate, but also to modify a whole clause and as such have additional function as discourse particles. In discourse function these particles are placed in clause-initial position; in direct negation function they occur adjacent to the clause constituent that they negate. The full inventory of discourse particles is:

kari		'a contrasting event' (gloss 'but') (see §5.5.3)
yamba/yambada/yambala		'a contrasting state' (gloss 'but')
kaki		'when, if'
kunka		'in case' (specified hypothetical event)
nganganjirrka/nganganjida		'in case' (not necessarily specified hypothetical event)
balu		'wrong assumption' (gloss 'wrong')21
nguba		'maybe, perhaps'

(Like glosses of other discourse particles, e.g. kunka 'in case', glosses for kari and balu in discourse function are represented by appropriate English lexical items, rather than by abstract terms, to distinguish their discourse function from their negation function.)

Kari, yamba, kaki, kunka and nganganjirrka function as co-ordination particles, although ngangnanjirrka may also occur independently outside a clausal context, and are explained in §4.3.2. For syntactic behaviour of balu and nguba see §4.9.2 and §4.9.3 respectively.

²¹ A similar 'mistaken thought' particle can be found in the Western Desert language Ngaanyatjarra (Glass 1 980).

1 1 4 Chapter 3

3.10 Clitics

There are several bound morphemes in Kuku Yalanji which always occur word-finally, i.e. after roots, stems or inflected forms. They may be categorised as enclitics rather than post-inflectional suffixes, because (a) they cannot be stressed, and (b) their co-occurrence is not confined to particular parts of speech, but they may attach to a variety of host words.22 Kuku Yalanji enclitics are grouped below according to their function, including descriptions of their use and/or reference to following sections where their functions are discussed more fully.

3. 10.1 Temporal clitics

-da	'then' (following action or state)23 (TEMP1)
-ku	'before' (preceding action or state)
-kuda	'meanwhile' (?) (TEMP2)
-ngVrr/V ~ -Vrr/C-	'first of all' (TEMP3)
wawu-	'since, after' in connection with Ablative (TEMP4)

Clitic -da is most frequently used with verbs, but has also been attested with all other parts of speech. It typically occurs on the last of a sequence of verbs to indicate that this is the final in a sequence of actions, for example:

(209) Jana kada-ny, kuyu mani-ny, waju-ny, nuka-ny-da. 3pl.NOM(S) come-PAST fish.ABS(O) get-PAST cook-PAST eat-PAST-TEMP1 'They came, caught a fish, cooked [it], and then ate [it].'

(For discourse function of clitic -da in temporal cohesion see §5.2.2; for use with imperatives see §4.7.)

The semantic content of clitic -ku is sometimes ambiguous and a distinction between temporal and emphatic -ku (see §3.10.2 below) is often impossible. (Note that -ku also functions as adverb marker and is homophonous with one allomorph of dative inflection (see §3.2.2).) Only in two time words does -ku clearly have a reference to times past, i.e. ngadiku 'a long time ago' and yilayku 'yesterday'. But while only some occurrences in discourse may be interpreted as having temporal reference, an interpretation as emphatic clitic is always possible. For this reason, and also because the cognate post-inflectional suffix -:gu in Guugu Yirnidhirr has clearly emphatic function (Haviland 1 979a:60), I favour this interpretation of -ku as emphatic marker.

Clitic -kuda, which is used mostly with verbs but also with nominals and pronouns, is not easily definable. Its most plausible interpretation as 'meanwhile' is explained in §5.2.2.24 If -ku is, or was, a temporal clitic meaning 'before', a combination of -ku + -da 'before then'

²² J.L. Klavans (1980) uses the term "host" for the word to which a clitic is attached.

²³ R. Hershberger (1964d) uses glosses -ku 'then' and -da 'now' in reference to a “prior time” and “now or following" time respectively. While on the whole my observations agree with hers, the status of -ku as temporal or emphatic clitic is ambiguous, as will be shown in this and the following section. See also H. Hershberger and R. Hershberger ( 1 982:294) "ku shows emphasis".

²⁴ R. Hershberger (1979:72 and 74) attested kuda as a free-form particle functioning as clause-initial question marker. I have heard this in conversation, but in elicitation this use of kuda was treated with misgivings.

may perhaps refer to a time span outside or 'alongside' this temporal symmetry, i.e. 'meanwhile'.

Clitic -(ng)Vrr 'first of all' has been attested with verbs, nouns and pronouns. It indicated either that a certain action is (or has to be) the first in a sequence, or that the referent is the first to do it, or the first to be affected, e.g. duda-ny-arr 'pounded [it] first, then ...' wulmanangka-ngarr 'old man(ERG) was the first to ...'.

(210) Nyulu kuyu-ngurr mani-ny, ngawuya dama-ny-da. 3sg.NOM(A) fish.ABS(O)-TEMP3 get-PAST turtle.ABs(o) spear-PAST-TEMPI 'He caught a fish first, then speared a turtle.'

With suppletive imperative forms and with trisyllabic roots with imperative inflection this clitic has also been attested as simply -rr, e.g. daya-rr 'give it first', julurri-ka-rr 'wash it first'. (For a full discussion of the discourse use of -(ng)Vrr see §5.2.2; for use with imperatives see §4.7.)

Proclitic wawu- occurs only with ablative forms, the whole word meaning 'since, after X', e.g.wawu-yinya-muny 'after that', wawu-mayi-muny 'after the meal'.

3.10.2 Emphatic clitics

-ku	general emphasis: 'really' (?) (EMPH 1)
-lu (Yalanji)/-la (Nyungkul)	'pay attention' (EMPH2)
-bi	surprise: 'lo and behold' (EMPH3)

Clitic -ku is used copiously in discourse with all parts of speech and is used as a 'general emphasis' marker. English translations may resort to 'really' for -ku, e.g. yalbay-ku 'really big', jalbu-ku '(it was) really the woman' and wubul-ku 'really a lot', but in most cases this would be too strong, particularly since 'really' also implies 'high degree'.

Yalanji -lu and Nyungkul -la are attention-getting devices in discourse (see §5.4.2) and may attach to all parts of speech. Clitic -lu in Yalanji marks especially important events by particularly frequent occurrence in limited stretches of discourse (see §5.5.1.1).

Clitic -bi, which also occurs with all parts of speech, conveys an element of surprise, similar to -bi in Djabugay, which is even stressed (Patz 1991:286). For details of its use see §5.4.2.

Clitic -rrku/V- ~ Vrrku/C- 'totality, exclusive' (EMPH4) may also be grouped among emphatic clitics (note its ending -ku), but it has more restricted application than the others. It is only used with verbs, pronouns, and particles kari 'not' and yala 'like, same'. With verbs -(V)rrku implies that an action is carried out totally and thoroughly, for example:

• (211) Maral-angka kambijulurri-ny-arrku. girl-ERG:pt(A) clothes.ABS(O) wash-PAST-EMPH4 'The girl washed all the clothes thoroughly.'
• (21) 2) Bayan kida-rrku! house.ABS(O) sweep.IMP-EMPH4 'Sweep the whole house thoroughly!'

With pronouns -(V)rrku stresses that only the particular referent is involved, for example:

1 1 6 Chapter 3

• (2 1 3) Ngayu wawu balka-nka yunu-ndu-rrku. 1 sg.NOM(S) want speak-PURP 2sg-LOC-EMPH4 'I want to speak only to you.'
• (2 1 4) Bama mungka-dunju nyulu-rrku kudamundu person.ABS(S) hair-curly.ABS(S) 3sg.NOM(S)-EMPH4 afterwards

jiba-badi-ny.

feel sorry-PAST

'Only the Islander felt sorry afterwards.' (Many years ago two whites and a Torres Strait Islander ('curly hair') had assaulted and robbed a settler in the area.)

(2 1 5) Ngayu-rrku bundanda-y. 1 sg.NOM(S)-EMPH4 sit.REDUP-NONPAST 'I am living by myself.'

3. 10.3 Other clitics

Clitic -baja 'very/again' has dual function as a stem-forming suffix with adjectives (see §3.2.3.8 (ii)) and a post-inflectional clitic with verbs. When attached to adjectives -baja always occurs in conjunction with emphatic clitic -ku, meaning 'very adjective'. On verbs -baja can apparently not be followed by -ku.

With state verbs, as with adjectives, -baja expresses a fairly high degree of intensity, as in yinilmanybaja 'were quite scared' in text 3, line 43, or:

(2 1 6) Nyulu bambay-ma-l-baja. 3sg.NOM(S) sick-INCHO-NONPAST-DEGREE 'He is getting quite sick. '

With action and motion verbs, on the other hand, -baja means 'verb again' (and will be glossed as 'again' in this context), for example:

• (2 1 7) Bajabaja-ngka jinkalmu bayka-ny-baja. blue.tongue.lizard-ERG:pt(A) taipan.ABs(o) bite-PAST -again 'The blue tongue lizard bit the taipan again.' (Blue tongue lizards are said to be able to fight and kill taipans.)
• (2 1 8) N gayu wawu kada-nka-baja. 1 sg.NOM(S) want come-PURP-again 'I want to come back.'

An idiomatic use of -baja has to be noted: yundu-ku-baja 'it's up to you'.

Clitic -ji (Yalanji), -yijin (Nyungkul)²⁵ attaches only to pronouns and expresses 'X's turn, X does something in turn' which may or may not imply an obligation to perform a particular action, for example:

(219)	Ngayu	yunu	mani	daji-l,
	1sg.NOM(A)	2sg.POSS	money.ABS(O)	give-NONPAST

²⁵H. Hershberger and R. Hershberger ( 1 982) record the NyungkuI form of this clitic as -jilin.

yundu-ji ngayku mznya daya. 2sg.NOM(A)-in.tum I sg.POSS meat.ABS(O) give.IMP 'I give you money and you, in tum, give me meat. ' (fhis refers to an exchange of favours rather than a business transaction.)

(220) Monday yindu ngayu-ji school-mun-bu dunga-ny, Monday other I sg.NOM(S)-in tum school-mun-LOC go-PAST mayi waju-nku. food.ABS(O) cook-PURP

'Last Monday it was my tum to go to the school to cook food [i.e. school lunch].'

(22 1) Yundu-ji bunda-y, ngayu bubu kida-l. 2sg.NOM(S)-in tum sit-IMP I sg.NOM(A) ground.ABS(O) sweep-NONPAST 'It's your tum to sit down, I sweep the yard.' (lit. ground)

3. 10.4 Ordering of clitics

Clitics may co-occur in the order:

temporal clitic

• emphatic clitics EMPHl-3

-ji 'in turn'

The forms -baja as verbal clitic and -(V)rrku EMPH4 have not been attested in conjunction with other clitics. The three temporal clitics cannot co-occur, but emphatic clitics EMPHl-3 may be used together in the numerical order of the subscript: -ku (EMPHl) -lu (EMPH2) -bi (EMPH3)·

3.11 Interjections

Kuku Yalanji has four monosyllabic and seven polysyllabic interjections, which cannot take any affixes and make up a distinct intonation group. A list of these with a description of their functions is given in §4. 1 0.

3.12 Morphological adaptation of English loan words

English noun roots with any number of syllables usually take catalytic suffix -mun- before inflections or stem-forming suffixes, e.g. car-mun-du (LOC), school-mun-ku (DAT), motorbike-mun-ji (COMIT). However, the catalytic suffix is not obligatory and English loans have often been attested without it, e.g. toyota-nga (LOC) (text 3, line 4), Chrisangka (ERG:pt) (text 3, line 1 5), car-nga (LOC).

English adjectives are suffixed with -bala , which is a feature in Pidgin (< 'fellow' as in waybala 'white fellow, white man'), e.g. drunk-bala, brown-bala, busy-bala-baja-ku 'very busy'.

Like many other Australian languages, Kuku Yalanji borrows English verbs as nominal roots which are then verbalised by inchoative or causative derivations. Intransitive verbs take the inchoative verbaliser -ma-l. If the root is monosyllabic, the inchoative suffix is usually

118 Chapter 3

reduplicated, e.g. travel-ma-l, shift-ma-I-ma-l. Note that the verb 'work' often takes the causative suffix -mani-l, used with intransitive verbs (see §3.8.5. 1 ), but in passivised form with suffix -ji-, e.g. ngayu work-ma-n-mani-ji-ny 'I worked (lit: I was made to work)'. Transitive verb roots are suffixed with -im²⁶ before combining with the transitive verbaliser -bunga-l ~ -ba-l, for instance: fix-im-bunga-l (fix-im-ba-l), push-im-bunga-l, count-imbunga-l, polish-im-bunga-l. (Note that only -bunga-l, and not -kanga-l, is used with English verbs.)

²⁶This suffix, presumably < 'him' as in 'fix him' is widely used in Australian pidgins and creoles (see Crowley & Rigsby 1 979).

4 Syntax

4.1 Simple sentences

A simple sentence in Kuku Yalanji normally consists of a verb complex (ve) and one or more noun phrases (NP) which fulfil particular syntactic and semantic functions in relation to the verb. But in the absence of copulas in Kuku Yalanji we have to distinguish between such verbal sentences and verbless or nominal sentences in which the verb slot remains unfilled.

To begin with, this section presents the possible constituents of noun phrases and of the verb complex, and their combinations. Sub-section §4.1 .3 discusses the syntactic types of verbs, followed by core-syntactic structures of simple sentences and optional extensions in §4. 1 .4. Nominal sentences are dealt with in §4. 1 .S and finally there are some remarks on the order of sentence constituents in §4.1 .6.

4. 1. 1 Noun phrases

A noun phrase in Kuku Yalanji may consist of several parts either in a continuous line or scattered in certain chunks throughout the sentence. All NP constituents have to agree in case marking whether the NP is continuous or discontinuous; only possessive constructions are exempt from this rule, in that case marking is not obligatory for the possessive form in a continuous NP (see §3.2.3.2 and §4.S.1). Continuous NPs where only the last constituent was marked for case have occasionally been encountered in discourse; however, when editing their own narratives, informants always corrected such occurrences to include case marking on all constituents.

A noun phrase may include the following constituents:

(a) A proper noun, for example:

Mandu, Jinabaji, Mabel	(personal names)
Wangkabaja, Bibikarrbi	(place names )

(b) A common specific noun, for example:

maral	'girl'
kalnga	'mother's brother'
kundurr	'death adder'
wada	'wild apple'

(c) A generic term (see §3 .1.2 (i)) on its own or preceding (b), for example:

jarba	'snake'
mayi wukay	'plant food hairy yam'
minya ngangkin	'flesh food porcupine'
juku buyku	'tree paperbark'

The use of generic nouns as a kind of classifier together with specific nouns is quite common in Yidiny and Guugu Yimidhirr. In Kuku Yalanji, on the other hand, the only generic terms that are used frequently in a classifying function are minya 'flesh food' (in contrast to inedible animals) and mayi 'plant food' (in contrast to inedible plants). Other generic terms have only occasionally been attested as classifiers in conjunction with specific nouns.

(d) A demonstrative pronoun (see §3.7) on its own or together with (b) or (c) or both, for example:

yinya	'that'
yinya karrkay/karrkay yinya	'that child'
yanyu dikal (wakuka)	'this bird (kookaburra)'

In combination with one noun the demonstrative may precede or follow the noun; ill combination with two nouns the demonstrative is preferred in initial position.

(e) An interrogative pronoun (see §3.6. 1) on its own or together with (b) or (c), for example:

wanya		'who'
wanya warru		'which young man'
wanyu minya		'what meat'

(f) A personal pronoun (see §3.5. 1) on its own or followed by any of (a}-(d), for example:

yundu	'you (sg)'
yubal Coraleen	'you and Coraleen'
bula ngamu	'the two mothers'
jana bama	'the Aborigines'
nyulu yinya	'that one'

The anaphoric and/or definite reference of noun phrases of this type is discussed in §5.2.1 .1 and §5.2.1 .2.

(g) One or more modifiers such as adjectives, comitative or privative forms in attributive function and quantifiers may be optionally included in a noun phrase of type (b), (c), (d) or (f), for example:

dubal murru-murru	(bark rough)	'rough bark'
jambul kaya kuli-ji	(two dog rage-COMIT)	'two vicious dogs'
yinya yalbay junka-ji	(that big strength-COMIT)	'that big and strong one'
nyulu kiru-kari	(3sg brain-PRIV)	'the stupid one'

The preferred order is for the modifier to follow the head constituent, but where an adjective and a quantifier are used together the quantifier tends to precede and the adjective to follow the head. Combinations of two or even more descriptive adjectives are rare.

(h) A nominal or pronominal possessive form may precede or follow a noun. An optional modifier (g) may be included after the head noun, for example:

badur ngamu-mu	(fishing.line mother-POSS)	'mother's fishing line'
ngayku bayan (yalbay)	(1sg.POSS house big)	'my (big) house'
kalka (jirakal) dingkar-amu	(spear new man-POSS)	'the man's (new) spear'

(i) An inalienably possessed part may stand in apposition to the head noun (see §4.5.2), for example:

nyulu dukul 'he [his] head'

The possessed part always follows the possessor; optional modifiers refer to the thing possessed, for example:

nyulu dukul yalbay 'he [his] big head'

Noun phrases of the last type may be analysed as consisting of two heads in the relationship of "inclusion apposition" as defined by Quirk et al. (1 972:637). It refers to "cases of apposition where the reference of the first appositive is not identical with that of the second, but rather includes it". The same analysis can be applied to generic/specific noun NPs (type (c)) and pronoun/noun NPs where the pronoun implies a greater number of referents than the noun as in yubal Coraleen 'you two (one of you) Coraleen (i.e. you and Coraleen)" or jana Ivy 'they (one of them) Ivy (i.e. the Ivy mob)'. 1

4. 1.2 The verb complex

The core constituent of the verb complex is a simple or compound verb such as kida-l 'scrape', nyunba-y 'tum back', warngku-wuna-y 'sleep', maku-nyaji-l 'meet'. The verb may be accompanied by a non-inflecting modifier which normally precedes the verb but may also follow it. Verbal modifiers can be:

(a) an adverbial particle (see §3 .8.6.1 ), for example:

kanban kuni-l	'hit by mistake'
walku dunga-y	'walk past'

(b) a derived adverb (see §3.8.6.2), for example:

kiray-ku wukurri-l	(secret-ADV follow)	'follow secretly'
kuli-ji-ku kuni-l	(rage-COMIT-ADV hit)	'hit viciously'

Verb complexes with more than one modifier have not been attested.

4. 1.3 Verbs - syntactic types

Kuku Yalanji distinguishes strictly between tranSItive and intransitive verbs. An intransitive verb requires only a noun phrase in subject function, which can be either a pronoun in nominative case or a nominal in absolutive case. The subject of an intransitive verb may be either animate or inanimate.

See also the interpretation of "apposition noun phrase" by R. Hershberger ( 1 964a:3 8).

1 22 Chapter 4

A transitive verb has to be accompanied by two core syntactic noun phrases. These are one noun phrase in agent function, which may be a pronoun in nominative case or a nominal in ergative case, and one noun phrase in object function, which may be a pronoun in accusative case or a nominal in absolutive case. In a context-related sentence either of the core syntactic noun phrases may be omitted, leaving the verb as sole clause constituent (e.g. text 36, lines 39, 42 and 45).

In Kuku Yalanji the transitivity value of a verb in a transitive sentence is not affected according to whether the action is performed by an animate or inanimate actor.2 An inanimate actor receives instrumental case marking to indicate that the action is uncontrolled and non-volitional as in examples (222) and (223). Chance actions by potentially controlling agents, on the other hand, may be expressed by passive constructions involving an intransitivised verb (see §4.2.4.3).

• (222) Karrkay mili-bu child.ABS(O) stinging. tree-INST 'A stinging tree burnt the child. ' waJu-ny. bum-PAST
• (223) Nganya juku-bu baka-ny. I sg.ACC(O) stick-INST poke-PAST 'A stick poked me.' (e.g. when I sat on the ground)

Because 'neutral' ergative and instrumental are identical in form, the syntactic function of the inanimate actor in such a sentence is ambiguous on first sight. This problem is considered further in §4.1 .4.4 (i).

4. 1.4 Case functions

In §3.2.2 the forms of case inflections were presented with only brief indication of their functions. This section will examine in detail the syntactic and semantic functions of cases in Kuku Yalanji.

4. 1.4.1 Preliminary remarks

This study is mainly concerned with providing a pragmatic formulation of Kuku Yalanji grammar and therefore uses the overt case form as a basis for the definition of 'case'. An analysis based solely on case roles according to syntactic-semantic relationships3 would appear impracticable before formal characteristics of Kuku Yalanji cases have been established. However, while using this starting point, I also pay close attention to case functions and will distinguish two cases such as Ergative and Instrumental, although they have the same form, because there are pressing syntactic reasons for doing so. On the other hand, there is the necessity to distinguish four core syntactic cases on formal grounds: ergative and absolutive for nouns and nominative and accusative for pronouns. The functions of these overlap in the following way:

3 See for instance Fillmore ( 1 968); A. Hale ( 1973); Foley (1980).

² A similar situation obtains in Guugu Yimidhirr (Haviland 1 979a: 125). In Yidiny (Dixon 1 977a:275) and Djabugay (patz 1991 :297-298) an inanimate agent requires a verb with derivational suffix -:di-n and -yi respectively, but the verb remains transitive.

Such a split case-marking system is a common phenomenon in Australian languages and it has become a firmly established convention to distinguish the functions of the respective cases as:

• A transitive subject (or Agent)
• S intransitive subject
• o transitive object

(see Dixon 1 972:xxii; Silverstein 1 976: 1 1 2). Since the case forms of nouns and pronouns do not correspond with these functions in a one-to-one relationship, one cannot simply adopt the functional labels as case labels, even though the functions, and not the cases, are of primary syntactic importance. I therefore gloss core syntactic nouns and pronouns with both the case and functional label; e.g. NOM(S) denotes a nominative pronoun in intransitive subject function, ABS(O) a noun in absolutive case and in transitive object function, etc. This may appear cumbersome, but I regard it as necessary for the following reasons: (a) all constituents of a noun phrase must have the same case function; and (b) only noun phrases with the same case function may be conjoined. If nominal and pronominal constituents of the same noun phrase or of conjoined noun phrases are glossed only with their respective case labels the identical function cannot be obvious, but the functional terms A, s and 0 will disambiguate the syntactic structure.

4. 1.4.2 Semantic functions of case inflections

Some Australian languages have noun classes which are indicated either by nominal prefixes, as in the languages of Arnhem Land and the Kimberleys, or by free-form determiners as in Dyirbal. Noun classes may relate to gender, animacy and edibility of plants or animals, among other semantic criteria. Kuku Yalanji employs a double set of inflections for ergative, dative, general locative and ablative, which are used, roughly, in accordance with animacy or inanimacy of the referent. H. Hershberger (1 964b:74) assumed that there are in fact two noun classes, "animate" and "inanimate" which determine the choice of the appropriate case inflection. However, further research has shown that the two sets of case inflections do not correspond with two fixed classes of 'animate' and 'inanimate' nouns. It rather appears that it is not an inherent and unchangeable feature of the noun that determines the inflection, but that it is the inflection which adds a semantic aspect (other than a case role) to the noun.

It is not all that unusual for inflections to combine the function of case marker with a non-syntactic function, such as indicating number (Latin) or definiteness (Turkish). However, a system of portmanteau case markers which effect a certain interpretation as to

1 24 Chapter 4

the 'animacy' of the inflected noun appears to be quite rare. One other somewhat similar example among Australian languages could be Mangarayi, a Western Roper River language, which uses prefixes that indicate both noun class and case (Merlan 1 982).

To determine the 'extra-grammatical' functions of Kuku Yalanji case inflections and the semantic basis for these functions we first of all have to consider the animacy hierarchy of referents and the distribution of the different types of inflections. A wide range of nouns around the mid-section of the animacy hierarchy will accept case markers from either set which indicates that the rigid terms 'animate' and 'inanimate' may be inappropriate for the two sets of inflections. For the time being I will therefore refer to the two inflection sets as set X and set Y. The animacy hierarchy for Kuku Yalanji nouns and their respective case marking is illustrated in Table 4. 1.

	Nominal referent	Case marking
(1)	humans (including deceased persons and most kinship terms) personified mythical beings ghosts and spirits dogs	set X
(2)	generic terms with animate reference4 vertebrates invertebrates natural forces (sun, wind, electricity, etc.)	set X or set Y
(3)	plants food geographical/geological features body parts, dead bodies language, illness, ceremonies some kinship terms5	set Y

Table 4.1: Animacy hierarchy and case marking

To investigate more fully the use of case inflections in Kuku Yalanji we also have to consider the case marking used for lexical items other than nouns, i.e. pronouns, adjectives, demonstratives, interrogatives, possessive forms, numerals, and location and time words. The case-marking categories for these parts of speech are listed in Table 4.2.

Pronouns, which take regular nominal case inflections for all but the core syntactic cases, generally refer to beings high up on the animacy scale, corresponding with those in category (1) of Table 4. 1. First, second and third person pronouns all behave in the same way.

⁴ For a list of generic terms see §3. 1 .2 (i).

⁵ A full list of these kinship terms can be found in §3.2.4.

NP constituent	Case marking
pronoun	set X
adjective	set X or set Y
demonstrative	set X or set Y
interrogative: 'who'	set X
'what', 'where'	set Y
possessive noun or pronoun form	set Y
numeral	set Y
location and time words	set Y

Table 4.2: Case marking on noun phrase constituents other than head nouns

Adjectives and demonstratives agree in type of inflection with the head noun of their NP. If they occur by themselves, in an anaphoric situation, their inflectional type still agrees with the understood head noun.

Numerals, although they have a similar function to adjectives in that they provide an additional comment on a noun, do not, as it were, describe an inherent characteristic of the referent. They just add the 'external' aspect of quantity which is variable for any referent. This may be the reason why their type of case marking is not bound to that of a co-occurring noun. A similar aspect of only 'remote link' to the head noun applies for possessive forms, which are also independent in their type of case inflection.

Location and time words, which take nominal case inflections from the set of local cases, clearly stand outside the animacy hierarchy.

Two points emerge from the distribution of type X and type Y case inflections among nouns and other noun phrase constituents:

• (a) For some NP-constituents there is no choice between the types of case inflections. Those that have to take type X inflections have referents high up on the animacy scale (human nouns etc., pronouns, interrogative 'who'). Those that have to take type Y inflections either have referents low down on the animacy scale (plants, tools, etc., interrogative 'what') or may be regarded as standing outside the animacy hierarchy, i.e. possessive forms, numerals and location and time words. (However, why some kinship terms (see §3.2.4) should be included here is quite unclear.)
• (b) If we have a dichotomy of high ranking 'animacy' on one extreme of the scale and 'inanimacy' or 'neutrality' on the other extreme, the choice of inflectional type in the middle section (set (2) in Table 4. 1) must be linked to some aspect of animacy. What is this aspect and how can the two inflection sets be identified accordingly?

Since numerals, possessive forms and location and time words which occupy a neutral position outside the animacy hierarchy invariably take set Y case markers, I call set Y inflections 'neutral'. If these inflections can be regarded as unmarked for semantic content, such content must be looked for in set X inflections.

The feature 'animate' is clearly inaccurate for set X since, as is shown in category (2) of . Table 4. 1, taxonomically 'animate' referents may also take neutral case marking. Consider the difference in case marking for sentences of the following kind. (For brevity I will only give English translations here and indicate the appropriate type of case marking; more examples in Kuku Yalanji will be found in sections §4.1 .4.3 and §4. 1 .4.4.)

,------------_ .. _-------

• (i) a. I sat on a mosquito (mosquito + neutral locative) b. I was stung by a mosquito (mosquito + set X ergative)
• (ii) a. I like mudcrab (mudcrab + neutral dative) b. I am afraid of mudcrabs (mudcrab + set X dative)
• (iii) a. A brown snake bit me (brown snake + neutral ergative)
  • b. The brown snake is eating a rat (brown snake + set X ergative)

All a. sentences show the nominal referent in an inactive or involuntarily active state, while all b. sentences mark the referent as a voluntary or potential actor. Neutral case marking in (ii)a. interprets the mudcrab as food while set X inflection in (ii)b. refers to the mudcrab as a potentially threatening actor with its propensity for nipping people's extrerrutles. Neutral case marking in (iii)a. interprets the brown snake bite as an unpremeditated reflex action on provocation while set X inflection in (iii)b implies that the snake is in full possession of its faculties when it feeds on a rat. One could argue that a mudcrab also nips only when trodden on or otherwise disturbed, and that there is therefore inconsistency in the case marking for (ii)b. and (iii)a. However, one important aspect of the use of the two inflection sets in Kuku Yalanji is that, where a choice is possible, a speaker may exercise this choice according to their own interpretation. Thus, while speakers would mostly coincide in their intuitions, one may encounter different perceptions as to the potential for voluntary or involuntary action of a certain referent. For instance 'a brown snake bit me' has also been attested with set X ergative marking. It could be that when talking about a potential threat like the mudcrab in (ii)b., emphasis is laid on the beast's ability for action, while the main characteristic of an accomplished snake bite is seen by most speakers to be an unpremeditated reflex.

Consider also:

• (iv) a. They lie in the sun (sun + neutral Locative)
  • b. The sun dries the ground (sun + set X Ergative)

Many speakers here credit the sun with an internal power with which it can act unprompted and unprovoked as in (iv)b. (There is no indication that the sun is personified in Kuku Yalanji, as it is in some other Australian languages.)

The fact that animate generic terms may take either type of inflection could be seen as underlining the abstract content of such terms, i.e. that they refer to concepts rather than actual referents. The choice of type X inflection with a generic term can be interpreted as indicating a 'real' referent, whereas 'neutral' inflection refers to the abstract concept. Consider:

• (224) a. Dingkar-angka karrkay . kuni-ny. male/man-ERG:pt(A) child.ABS(O) hit-PAST 'That was a man what [sic] hit the child. ' (not a woman; I saw him)
  • b. Dingkar-abu karrkay kuni-ny male/man-ERG(A)

'Some man hit the child.' (I think it was a man; but it could've been another child).

However, neutral inflections with human generic terms are very rare, whereas they become more frequent as one proceeds lower down in the animacy scale.

The semantic clues gleaned from sentences (i) to (iv) fairly well circumscribe the semantic content of set X case inflections. It seems that the most appropriate term for this semantic content is "potent", following Chafe's use of the term in his specification of nouns. "Potent" as defined by Chafe (1 970:1 09) refers to a nominal referent that "has, or is conceived to have, its own internal power" (my emphasis). The notable fact in Kuku Yalanji is that this semantic feature does not have to be inherent in the noun (although it often is, as in the nouns in category (1), Table 4.1 ), but is added to the noun by the case inflection.

To summarise, the two sets of non-zero case inflections in Kuku Yalanji can be termed 'neutral' and 'potent' according to their semantic content. For a large range of nouns there is a choice between 'neutral' and 'potent' case inflections and speakers can exercise this choice according to what they want to express.

To distinguish nouns that always take only one of the two possible inflections from those for which there is a choice I refer to the three categories of Table 4.1 :

category 1	('potent' inflection only)
category 2	('potent' or 'neutral' inflection)
category 3	('neutral' inflection only).

4. 1.4.3 Core syntactic cases

Core syntactic components of a sentence are those NPs which obligatorily occur with the verb according to its transitivity value.

An intransitive core has the following components:

NP (S) + intransitive vc.

A transitive core consists of:

NP (A) + NP (0) + transitive vc.

(For defintions of S, A and 0 see §4.1 .4. 1 .)

Nominal constituents of an NP in S function take absolutive case marking, which is zero, and pronominal constituents are in nominative case. (For pronoun forms see §3.5.1 ; for NP constituents see §4. 1 . 1 .)

An NP in S function denotes the performer of an action that does not affect an object as expressed by verbs of motion and some verbs of utterance; it may also refer to the experiencer with verbs of rest, bodily functions, emotion or physical state.6 Except for obvious semantic constraints imposed by the verb an NP in S function may be animate or inanimate.

• (225) Yinya kadar that.ABS(S) wallaby.ABS(S) 'That wallaby is running. ' warri-y. run-NONPAST
• (226) Kaba kada-y. rain.ABS(S) come-NONPAST 'Rain is coming. ' (standard conversation opener if there are any clouds in sight)

⁶ In my description of core syntactic functions I employ as far as possible terms used by Grimes ( 1975); I do not, however, use his term 'Agent' which combines some transitive and intransitive subject functions; 'Agent' in Australian case grammar refers exclusively to transitive subject function.

1 28 Chapter 4

• (227) Bula wulbuman 2du.NOM(S) old.woman.ABS(S) 'The two old women sang out. ' yirrka-ny. sing. out-PAST
• (228) Nganjan wunana-y. father.ABS(S) lie.REDUP-NONPAST 'Father is resting.'
• (229) Juku wubul yinyay janjana-y. tree.ABS(S) many.ABS(S) there stand.REDUP-NONPAST 'Many trees are standing there. '
• (230) Nyulu karrkay kalji-l. 3sg.NOM(S) child.ABS(S) vomit-NONPAST 'The child is vomiting. '
• (23 1) Nyulu bunjil jibabadi-y. 3sg.NOM(S) widow(er).ABS(S) feel.sad-NONPAST 'The widow[er] is sad. '
• (232) Ngayu wawula-y. « wawu-wula-y) 1 sg.NOM(S) breath.die-NONP AST 'I am exhausted. '

Nominal constituents of a NP in A function take ergative case marking (see §3.2.2) and pronominal constituents are in nominative case as for S function (see §3.5.1 ).

A NP in A function refers to the performer of an affecting or effecting action on an object, or the experiencer of a sensory perception such as 'see', 'hear', 'smell'. The action may be intentional or accidental, but it must be performed by a referent of a category 1 or category 2 noun, that is someone or something which is thought to be able to act of hislher own accord. (An accidental action by a category 1 noun may be expressed by the adverb kanban 'by mistake' in a normal unmodified transitive sentence, or by a passive construction. See §4.2.4.2 and §4.2.4.3 for details on accidental action.)

Category 2 nouns in A function take 'potent' inflection if credited with responsibility for their action and 'neutral' inflection if not. (Since 'neutral' ergative forms are the same as instrumental forms it is sometimes ambiguous whether a category 2 noun is in ergative or instrumental case. Distinguishing features between ergative and instrumental are discussed in §4. 1 .4.4 (i).)

Examples of category 1 nouns and pronouns in A function:

• (233) Nyulu warru-ngku ngawuya dama-ny. 3sg.NOM(A) yg.man-ERG:pt(A) sea.turtle.ABS(O) spear-PAST 'The young man speared a sea turtle.'
• (234) Bula jalbu-bulal-angka bayan ngara-ny. 3du.NOM(A) woman-pair-ERG:pt(A) camp.ABS(O) set.up-PAST 'The pair of women set up camp. '
• (235) Dubu-ngku nyunguny makunyaji-ny karrkay. ghost-ERG:pt(A) 3sg.ACC(O) meet-PAST child.ABS(O) 'The ghost met the child. '

(236) Dingar-angka kaya kanban kuni-ny. man-ERG:pt(A) dog.ABS(O) by. mistake hit-PAST 'The man hit the dog by mistake.'

Examples of category 2 nouns in A function:

(237)	Yinya-ngka	kubarr-angka	yalbay-ngka	maral	bayka-ny.
	that-ERG:pt(A)	eel-ERG:pt(A)	big-ERG:pt(A)	girl.ABS(O)	bite-PAST
	'That big eel bit the girl.'

• (238) Nganya bambay-bunga-ny kubarr-da. I sg.ACC(O) sick-CAUS-PAST eel-ERG(A) 'The eel [meat] made me sick. '
• (239) M alal-angka kumu spider-ERG:pt(A) mosquito.ABS(O) 'The spider grabbed the mosquito. ' karrba-ny. grab-PAsT
• (240) N ganya murraja-mun-du I sg.ACC(O) stonefish-mun-ERG(A) 'A stonefish poked me.' baka-ny. poke-PAsT
• (241) Nyunguny mili-bu 3sg.ACC(O) jellyfish-ERG(A) 'A jellyfish burnt himlher.' waju-ny. burn-PAST

Nominal constituents of a NP in 0 function stand in absolutive case, as for S function, and pronominal constituents are marked for accusative. Examples of nouns in 0 function can be found in sentences (233), (234), (236), (237); examples of pronouns in 0 function are in (240), (241 ).

A NP in 0 function denotes a patient which is affected by an action (examples in (233), (236), (237), (238), (239), (240) and (24 1)), or factitive which is the object that is effected by an action (example in (234)), or an experiencer (as in sentence (235)).

4. 1.4.4 Peripheral syntactic and local cases

Core sentences in Kuku Yalanji may be extended by one or more NPs in a peripheral syntactic or local case. These are: instrumental, dative, general locative, ablative, abessive and perlative.7 (For inflectional forms see §3.2.2.) In my identification of the semantic roles of these cases I follow Blake (1 977).

(i) Instrumental

The instrumental case indicates the implement, which may be a body part, with which an action is performed. There is no 'potent' case inflection for instrumental, which shows that the implement employed in an action is in no way thought of as having any 'own internal power'. Examples:

⁷ See R. Hershberger (1 979) for NP constituents in instrumental case (pp.50-5 1), associative (dative) (pp.5 1-52) and locative including perlative, which she analysed as "instrument case marker + _rr ⁿ (pp.52-54).

1 30 Chapter 4

• (242) Dingar-angka yawu dama-ny yinba-bu. man-ERG:pt(A) stingray.ABS(O) spear-PAST 3-prong spear-INST 'The man speared the stingray with a three-pronged spear. '
• (243) N gayu mayi yala-ku dingka-l mara-bu. I sg.NOM(A) fruit.ABS(O) thiS-ADV squeeze-NONPAST hand-INST 'I squeeze the fruit like this with my hand.'

If the instrument should be an otherwise animate noun, which has never been encountered spontaneously but only after a lot of prompting, it appears that the 'instrument' is perceived to be the inanimate body of the referent,8 for example:

(244) Warru-ngku kaya jarba-bu kanga-ny. yg.man-ERG:pt(A) dog.ABS(O) snake-INST chase-PAST 'The young man chased the dog with a snake.' (i.e. throwing a dead snake at it)

Concepts like 'I hunt pigs with a dog' where the 'instrument' is an active participant accompanying the agent are expressed by comitative (see §4.6. 1 ).

Instrumental case inflections coincide in form with the 'neutral' ergative inflection, but there are syntactic reasons for distinguishing the two cases:

(a) A noun in instrumental case may occur in an intransitive sentence whereas this is impossible for a noun in ergative case, for example:

• (245) Jana jalbu jina-bu dunga-y. 3pl.NOM(S) woman.ABS(S) foot-INST go-NONPAST 'The women are walking on foot. ' (lit. going with feet)
• (246) Maral bungku-bu janjana-y. girl.ABS(S) knee-INST stand.REDUP-NONPAST 'The girl is kneeling. ' (lit. standing with knees)

(b) An instrumental NP remains unchanged in a reflexive construction (see §4.2.4.2) whereas an ergative NP changes to absolutive case (see Blake 1 977:45).

• (247) a. Ngayku-wun-du ngamu-ngku minya yaka-l naybu-bu. I sg.POSS-ERG(A) mother-ERG:ptCA) meat.ABS(O) cut-NONPAST knife-lNSI 'My mother is cutting meat with a knife.'
  • b. N gayku ngamu yaka-ji-ny 1 sg.POSS.ABS(S) mother .ABS(S) cut -REF-PAST 'My mother cut herself with a knife.' naybu-bu. knife-INST

(c) An instrumental NP remains unchanged in a passive construction, which involves the syntactic change A> LOC and 0> S (see §4.2.4.3). Passivisation of sentences containing an NP with instrumental/'neutral' ergative inflection shows that firstly, a category 3 noun is always understood to be in instrumental case even where there is no explicit agent, and secondly, it may be ambiguous whether a category 2 noun is in instrumental or ergative case. Consider the following:

(248) a. Nganya juku-bu kuni-ny. I sg.ACC(O) tree-INST hit-PAST 'A tree hit me.'

⁸ See Fillmore as referred to in Blake (1 977:47).

b. Passive (0 > S)

1----

Ngayu juku-bu kuni-ji-ny. I sg.NOM(S) tree-INST hit-ITR-PAST 'I was hit withlby a tree.'

• (249) a. Nyunguny mili-bu 3sg.ACC(O) stinging.tree-INST burn-PAST 'A stinging tree burnt him.' waJu-ny.
  • b. Passive (O > S)

Nyulu mili-bu waju-ji-ny. 3sg.NOM(S) stinging.tree-INST burn-ITR-PAST 'He was burnt withlby a stinging tree. '

If there were a human or some other controlling agent responsible for the events in sentences (248) and (249), this agent would be stated and would appear in 'potent' ergative case in the transitive clause and in locative case in the passive clause. In the absence of an explicit agent the tree, or branch, in sentence (248) and the stinging tree in (249) are apparently still regarded as mere instruments of some force which, however, cannot be stated since it is not known. According to the 'chance' aspect of such agentless events the passive construction, which topicalises the patient, seems to be preferred by Kuku Yalanji speakers.

Now consider the passivisation of sentences (240) and (241) into (250) and (25 1) respectively:

• (250) N gayu murraja-mun-du / murraja-nga baka-ji-ny. 1sg.NOM(S) stonefish-mun-INST / stonefish-LOC poke-ITR-PAST 'I was poked by a stonefish.'
• (25 1) Nyulu mili-bu / mili-nga waju-ji-ny. 3sg.NOM(S) jellyfish-INST / jellyfish-LOC burn-ITR-PAST 'He was burnt by a jellyfish. '9

and:

(252) a. Jarba-bu nyunguny bayka-ny. snake-ERGIINST 3sg.ACC(O) bite-PAST 'A snake bit him. '

b. Passive

Nyulu jarba-nga / arba-bu bayka-ji-ny. 3sg.NOM(S) snake-LOC / snake-INST bite-ITR-PAST 'He was bitten by a snake.'

In sentences (250}-(252) it is possible to change the NP referring to stonefish, jellyfish or snake to locative in the passive, which indicates underlying agentive function, or to retain instrumental inflection (a choice which is not possible in (248) and (249)). Therefore it is ambiguous whether the respective NP in the corresponding transitive sentence is in fact in ergative or instrumental case. Only speakers' preferences, which are indicated by the order in which the choice of NPs in passive constructions is presented, would imply that stonefish and jellyfish are more 'instrumental' and snake is more 'agentive'.

⁹ Mili is a homonym which refers to 'stinging tree' in (249) and to '(box)-jellyfish' in (241) and (251).

1 32 Chapter 4

(ii) Dative

A NP in dative case (for forms of inflections see §3.2.2) has basically two distinct functions. The most common one is to comment on the reason why the action mentioned in the sentence core is performed. In the second function a dative NP sets a parameter with respect to which a certain action or state can be seen. Category 2 nouns take 'potent' or 'neutral' case inflection according to whether the referent is regarded as an active animate being as in (256b) and (26 1 ), or as an inanimate substance such as potential food as in (256a).

The broadly termed function of 'reason' is twofold again. One aspect of it is goal of a purposeful action (but not of a purposeful movement, which is expressed by locative). This includes an indirect beneficiary, i.e. someone for whom an action is performed so that he/she does not have to do it him/herself.¹⁰ Complements of verbs like 'learn, teach' and the adjectival particle 'want/like'¹¹ may also be included under goal.¹² Examples:

• (253) N gayu bubu kida-I-kida-l bayan-ka. 1 sg.NOM(A) ground.ABs(o) sweep-l-REDUP-NONPAST camp-DAT 'I am sweeping the ground for a camp. '
• (254) Nyulu Cooktown dunga-y yaba-nka. 3sg.NOM(S) Cooktown go-NONPAST eB-DAT:pt 'He goes to Cooktown for his elder brother. '
• (255) Bama-ngka nganya binal-bunga-l kuku-ku Aboriginal-ERG:pt(A) 1 sg.ACC(O) know-CAUS-NONPAST word-DAT

mayi-ka minya-ka bubu-ku bama-nka. plant.food-DAT flesh.food-DAT land-DAT Aboriginal-DAT:pt 'Aborigines teach me about [their] language, "bush-tucker", land [and] Aboriginal people. '

• (256) a. Ngayu kulngu-ku wawu jirray. 1 sg.NOM(S) bandicoot-DAT like much 'I like bandicoot a lot.' (i.e. to eat)
  • b. N gayu kulngu-nku wawu 1 sg.NOM(S) bandicoot-DAT:pt like 'I like [my pet] bandicoot a lot.' jirray. much

The other aspect of 'reason' is the (often indirect) cause for an action or emotion. This includes a general threat which may evoke an emotion as in (26 1) or something for fear of which something is done or not done as in (262) and (263). (If the threat is a direct one the ablative case is used; see (iv) below.) The aspect of 'cause' also includes the complement of a verb of utterance such as 'talk about' or 'ask for/about' as in (264).

Examples:

(257) Nyulu dingkar ngamu-nku badi-ny. 3sg.NOM(S) man.ABS(S) mother-DAT:pt weep-PAST 'The man cried because of his mother. ' (she had died)

¹² See Grimes (1975:121) “range”.

¹⁰ See Beneficiary as defined by Blake (1977:35).

¹¹ For adjectival particles see §3.9.1.

• (258) Jana mayi nuka-l ngawa-nka. 3pl.NOM(A) food.ABS(O) eat-NONPAST baby-DAT:pt 'They are eating because of the baby.' (to celebrate the birth)
• (259) Ngali mayi-ka kuni-wa-y. 3du.exc.NOM(S) food-DAT hit-RECIP-NONPAST 'We two are fighting over food.'
• (260) N gayu wawurrwawurr-ma-ny bana-ka. I sg.NOM(S) glad-INcHO-PAST water-DAT 'I was glad about the water. ' (after a long dry walk)
• (26 1) Karrkay yinil kiju-nku. child.ABS(S) afraid crab-DAT:pt 'The child is afraid of crabs.' (they may nip)
• (262) N gayku babarr wawubaja-nga kari dunga-y bilngkumu-nku. I sg.POSS.ABS(S) eZ.ABS(S) river-LOC NEG go-NONPAST crocodile-DAT:pt 'My elder sister does not go to the river because of the crocodile.'
• (263) Ngayu wanarri-y kaya-nka kuli-ji-nka. I sg.NOM(S) run.REDUP-NONPAST dog-DAT:pt rage-COMIT-DAT 'I am running because of the vicious dog. '
• (264) Ngayu Mabel-anda babaji-ny bayan-ka jirakal-ka. 13 I sg.NOM(S) Mabel-LOc:pt ask.about-PAST house-DAT new-DAT 'I asked Mabel about her new house.'

(For syntactic details of constructions with this type of verb of utterance see §4.2.4.4.)

The parameter-setting function of dative is found in comparison-type constructions where a statement is made 'with respect to' a comparable or otherwise relevant factor. Examples:

• (265) N gawuya ngujay-ngka yalbay-bajaku. sea.turtle.ABS(S) freshwater.turtle-DAT big-very 'A sea turtle is a lot bigger than a freshwater turtle.' (i.e. compared to a freshwater turtle, a sea turtle is big)
• (266) Babi-ngka jarruka wubul baka-ny ngamu-nku. FM-ERG:pt(A) scrub.hen (egg).ABS(O) many'ABS(O) dig-PAST mother-DAT:pt 'Grandmother dug up a lot more scrub hen eggs than mother. ' (i.e. compared to mother, grandmother dug a lot).
• (267) Nyulu julmbanu yala maja bubu-ku 3sg.NOM(S) big.grey.kangaroo.ABS(S) like boss.ABS land-DAT

Australia-ka.

Australia-DAT

'The big grey kangaroo is like the boss of Australia. ' (i.e. with respect to Australia, the big grey kangaroo is like the boss)

¹³ An inanimate complement of such a verb of utterance may be in locative case, e.g. bayan-ba 'house-LOC' , but dative is preferred, presumably because locative is already used for the addressee.

134 Chapter 4

(iii) General locative

An NP in locative case (for forms of inflection see §3.2.2) has a variety of functions which may be divided broadly into four categories: 'goa1' (in a different sense than conveyed by dative), 'intermediary', 'locality' and the purely syntactic function of case marker for the underlying A or O NP in a passive or antipassive construction. A category 2 noun takes ‘potent’ inflection if the referent is perceived as an animate being as in (270) and (271), but 'neutra1' inflection where it is interpreted as an inanimate entity such as for example food or a mere locality as in (269) and (276).

Under the function of ‘goal’ we can list the goal of a purposeful movement and the addressee or receiver complement of a three-argument verb such as 'tell, show, ask, give'. Examples:

• (268) Nganjiny burrir-a mariji-ny. Ipl.exc.NOM(S) island-LOC swim-PAST 'We swam to the island.'
• (269) Warru-warru bikibiki-nga dunga-ny. yg.man-REDUP.ABS(S) pig-LOC go-PAST 'The young fellows went out for pigs.'
• (270) Charlie dunga-ny yarraman-anda. Charlie.ABS(S) gO-PAST horse-LOc:pt 'Charlie went to the horses.' (e.g. to feed them).
• (27 1) Ngayu minya milbi-l buji-nda. 1sg.NOM(A) flesh.food.ABS(O) show-NONPAST cat-LOC:pt 'I show the meat to the cat. '
• (272) Nyulu marrkin daji-ny yaba-nda. 3sg.NOM(A) gun.ABS(O) give-PAST eB-LOC:pt 'He gave the gun to his elder brother.'

A locative NP may denote the intermediary in the conveyance of something, for example:

(273) Ngayu yunu mayi yunga-l 1sg.NOM(A) 2sg.POSS.ABS(O) food.ABS(O) send-NONPAST ngayku-wun-bu ngamu-ndu. 1 sg-POSS-LOC mother-LOC:pt 'I will send your food with my mother.'

(274) Maja-ngka kuku yunga-ny buliman-anda. bOSS-ERG:pt word.ABS(O) send-PAST police-LOC:pt 'The boss sent a message with the policeman. '

Any transitive or intransitive clause may include an optional locative NP to indicate where the action is taking place. Locative in this function conveys 'in, on, at' a particular place; it also marks a noun co-occurring with a location word such as 'under, behind, on top'. With proper names of places such as mangkal ‘Cedar Bay’, and dikarra ‘Thompson Creek’, locative inflection is optional. Examples:

(275) Maral-maral bana-nga janji-l. girl-REDUP.ABS(S) water-LOC bathe-NONPAST 'The girls are bathing in the water.'

• (276) Charlie yarraman-ba bunda-yo Charlie.ABS(S) horse-LOC sit-NONPAST 'Charlie is sitting on a horse. ' ('horse' is seen as the location; cf. (270) where 'horses' show 'potent' inflection)
• (277) Dingar-angka kadar kuni-ny maja-nga. man-ERG:pt(A) wallaby.ABS(O) kill-PAST scrub-LOC 'The man killed a wallaby in the scrub.'
• (278) larba badabada juku-ngu wunana-y. snake.ABS(S) under log-LOC lie.REDUP-NONPAST 'A snake is lying under the log. '
• (279) Mary ngundu jana-y Queenie-nda. Mary.ABS(S) close.by stand-NONPAST Queenie-LOc:pt 'Mary is standing close to Queenie.'

The function of Locative in passive and antipassive constructions is discussed in §4.2.4.3 and §4.2.4.4.

(iv) Ablative

Ablative has the general function of implying a 'source', which in most of its manifestations coincides with a semantic 'former' (Grimes 1 975:133ff.). Specific aspects of this function are:

• (a) a locational or temporal starting point which has been left behind by the actor as in (280) and (28 1 );
• (b) the source of a possession as in (282);
• (c) the material from which something is made as in (283) and (284);
• (d) the cause of a physical state as in (285) (c.f. cause of emotion in dative case, e.g. (260), (26 1 » ;
• (e) a direct threat from which someone is trying to escape as in (286) and (287) (cf. indirect threat in dative case, e.g. (262), (263» .

A category 2 noun takes neutral inflection where it is interpreted as an inanimate body (284) and 'potent' inflection where it is regarded as a living and active being (287).

Examples:

• (280) Kija jambul-muny ngayu dunga-y-baja Cairns-muny. moon two-ABL I sg.NOM(S) gO-NONPAST-again Cairns-ABL 'After two months I will leave Cairns.' (lit. go from)
• (28 1) lalbu-ngku kaya kanga-l bayan-muny. woman-ERG:pt(A) dog.ABS(O) chase-NONPAST house-ABL 'The woman chases the dog from the house.'
• (282) labuju-ngku mayi kari nuka-l babarr-andamuny. yB-ERG:pt(A) food.ABS(O) NEG eat-NONPAST eZ-ABL:pt 'A younger brother must not eat food from [his] elder sister.'
• (283) Jalbu-ngku balji wukurri-l jilngan-muny. woman-ERG:pt(A) bag.ABS(O) weave-NONPAST reed.grass-ABL 'The woman is weaving a bag from reed grass.'
• (284) Dingkar-angka belt balka-ny kurriyala-muny. man-ERG:pt(A) belt.ABS(O) make-PAST carpet.snake-ABL 'The man made a belt from the carpet snake.'
• (285) Ngayu bam bay minya-muny buyun-muny. I sg.NOM(S) sick meat-ABL bad-ABL 'I am sick from the bad meat. '
• (286) Nyulu ngaki-ji-y 3sg.NOM(S) hide-REF-NONPAST police-ABL:pt 'He is hiding from the police.' buliman-andamuny.
• (287) Jukijuki warri-n-warri-y kurriyala-ndamuny. chicken.ABS(S) run-n-REDUP-NONP AST carpet.snake-ABL:pt 'The chickens are running from the carpet snake.'

(v) Abessive

A NP in abessive case refers to a place from which an action is performed without the actor leaving this location. Location words are used more frequently in abessive case then nominals. (For location words see §3 .3 .2.) There is no 'potent' inflection for abessive, possibly because a noun in this case can only be interpreted as a location and not as a potentially active animate being. Examples:

(288)	Mabel Mabel.ABS(S)	yirrka-y sing.out-NONPAST	bayan-mundu. house-ABESS
'Mabel is singing out from the house.' (cf. (281))
(289)	Jana	juku-mundu mayi	yilba-l.

3pl.NOM(A) tree-ABESS fruit.ABS(O) throw-NONPAST 'They are throwing the fruit from the tree.'

(vi) Perlative

An NP in perlative case describes passage through a defined space. Apparently the distinction between locative and perlative is made only in Kuku Nyungkul. Kuku Yalanji uses locative for both functions. There is no 'potent' perlative inflection. (In sentences like 'the spear went through the wallaby' locative is preferred for 'wallaby'; perlative is accepted only with some misgivings.) Examples: --------------------------------

• (290) Yira yangka wala-ny bujil-darr miyil-darr. JUIce green.ant.ABS(S) enter-PAST nose-PERL eye-PERL 'Green ant juice entered through the nose and through the eyes.' (having been inhaled for the treatment of a cold)
• (29 1) Jana wawubaja-burr mariji-ny. 3pl.NOM(S) river-PERL swim-PAST 'They swam through the river. '

(292) Dingkar baral-darr dunga-n-dunga-y. man.ABS(S) road-PERL go-n-REDUP-NONPAST 'The man is going along the road. '

A perlative NP may take the place of a locative NP in a passive or antipassive construction in the appropriate context (see §4.2.4.3 and §4.2.4.4.)

4. 1.5 Nominal sentences

,----

Since there is no copula in Kuku Yalanji various types of verbless or nominal sentences are used.14 A nominal sentence resembles an intransitive sentence in structure in that it has a quasi-intransitive subject in absolutive or nominative case. This is followed by a non-verbal predicate giving the structure:

[NP1]s non-verbal Predicate

meaning: S is Predicate. 1S

The distinction between subject and predicate in nominal sentences depends entirely on stress and intonation pattern. The subject is characterised by final rising intonation and may, in careful speech, be separated from the predicate by a pause. The normal stress pattern shows clause stress on the final constituent of the predicate, except in a comparison involving a dative NP as in (302). Division into subject and predicate is indicated by --; where different segmentations are possible according to intonation these are indicated by (--) with corresponding translations.

The predicate may be:

(a) A NP in absolutive case. In these constructions the subject NP often involves a demonstrative, for example:

• (293) Nyulu yinya ngayku 3sg.NOM(S) that.ABS(S) 1sg.POSS.ABS yZ.ABS 'That one is my younger sister. ' jinkurr.
• (294) lungkalu yinya -- kami-nga. billycan.ABS(S) that.ABS(S) MMIFF-POSS 'That billycan is grandmother's.'
• (295) Yinya (--) mayi (--) ngulkurr. that.ABS(S) food.ABS(S) good(.ABS) 'That is good food.' OR 'That food is good.'

(b) A NP in locative case or a location word, for example:

(296) Kami -- bayan-ba. MMIFF.ABS(S) house-LOC 'Granny is in the house. ' (i.e. at home)

¹⁴ See R. Hershberger (1979:70-72) 'Stative Clauses'. Note that the single constituent clauses included in her examples are only possible in anaphoric contexts where the subject NP is deleted. Note that her term "Benefactive" refers to my possessive.

¹⁵ The distinction between 'topic' and 'comment' (see Hockett 1958) would be particularly appropriate for such nominal sentences, but since I employ 'topic' as a discourse term (see §5.3) I rather use 'subject' and 'predicate' in this section.

1 38 Chapter 4

• (297) Jana karrkay -- badabada. 3pl.NOM(S) child.ABS(S) down. there 'The children are down there.'
• (c) A NP in ablative case, for example:
  • (298) Ngayku ngamu -- kuna-muny. I sg.POSS.ABS(S) mother.ABS(S) Shiptons.Flat-ABL 'My mother is from Shiptons Flat.'
  • (299) Yinya (--) kuyu (--) jalun-muny. that.ABS(S) fish.ABS(S) sea-ABL 'That is fish from the sea' OR 'that fish is from the sea.'

Note that while the ablative form in a nominal sentence may have attributive function, i.e. 'fish from the sea' = 'sea fish', it cannot be regarded as a derived adjective since it does not allow further case inflections.

• (d) An adjective or numeral, for example:
  • (300) Nyungu kalka -- dudu. 3sg.POSS.ABS(S) spear.ABS(S) blunt 'His spear is blunt.'
  • (301) Ngayku kangkal -- kulur. 1 sg.POSS.ABS(S) own.child.ABS(S) three 'I have three children. '

This type of nominal sentence may include an optional dative NP to indicate comparison as for instance (265) or:

(302) Ngayku kangkal -- yalbay nyungu-wun-ku kangkal-ka. 1sg.POSS.ABS(S) own.child.ABS(S) big 3sg-POSS-DAT own.child-DAT 'My child is bigger than her child. '

(e) A comitative construction which may convey ownership, accompaniment or a characteristic (see §4.6.1 ), for example:

(303)	Juku	ngakun --	nganka-ji.
		tree.ABS(S) flame.tree.ABS(S)	flower-COMIT
	'The flame tree has flowers.'

(304) Yundu maji-ji?16 2sg.NOM(S) match-COMIT 'Do you have any matches?'

(f) A privative construction which conveys the lack of something (see §4.6.2):

(305) Nyulu kiru-kari. 3sg.NOM(S) brain-PRIV 'He is stupid.'

¹⁶ In nominal sentences with only two words like (304) and (305) there is no intonation distinction other than clause stress between subject and predicate.

• (306) Nyulu wulman -- mungka-kari. 3sg.NOM(S) old.man.ABS(S) hair-PRIV 'The old man has no hair.'
• (g) Walu 'face, appearance' plus a nominal which refers to personal appearance:
  • (307) Nyulu -- walu 3sg.NOM(S) face 'He is ugly.' nganJay. ugly
  • (308) Bama -- walu ngumbu. Aborigine.ABS(S) face dark 'Aborigines are dark(-skinned). '
• (h) Yala 'like' plus a noun phrase to express an approximation:
  • (309) N gayu karrkay -- yala Eileen-anga kangkal. I sg.NOM(S) child.ABS(S) like Eileen-POSS-ABS own.child.ABS 'I was a child like Eileen's child (about the same age).'
  • (3 1 0) Mili-nga karra -- yala cotton. jellyfish-POSS.ABS(S) string.ABS(S) like cotton.ABS 'The jellyfish's tentacles are like cotton.' (so fine)

Yala may co-occur with walu in apparently any order (see §3.9.2 (i)):

(3 1 1) Yinya dingkar -- yala walu kadar. that.ABS(S) man.ABS(S) like face wallaby.ABs 'That man looks like a wallaby. '

(i) An adjectival particle plus an optional NP in dative case (for adjectival particles see §3.9. 1 ):

• (3 1 2) a. Maral yinil. girl.ABS(S) afraid 'The girl is afraid.'
  • b. Maral yinil jarba-nka. girl.ABS(S) afraid snake-DA T:pt 'The girl is afraid of snakes.'
• (3 1 3) Kaya wawu bajibay-ngka. dog.ABS(S) want bone-DAT 'The dog wants a bone.'

4. 1.6 Order of sentence constituents

Every NP in Kuku Yalanji is clearly identifiable for its role by case marking, and therefore the order of sentence constituents is of no great importance unlike, for instance, in English, where word order is the primary means of recognising syntactic roles. Word order is in fact quite free in Kuku Yalanji as in most other Australian languages, although there is a certain preferred order of sentence constituents. This is S V for intransitive and A O V for transitive sentences. Peripheral and local extensions usually follow the verb but may stand anywhere in the sentence. Where a dative or locative NP is an immediate complement of a three-argument verb, e.g. (264), (27 1) and (272), or of an adjectival particle, e.g. (256), (3 1 2b), (3 1 3), it will usually be placed adjacent to this constituent, either before or after.

S V and A 0 V may be regarded as the unmarked word order. A change in word order which moves a constituent other than A or S to the front serves to topicalise this constituent. (For discourse-related word order see §5.4. 1 .) A basic rule which tends to override the preferred word order is that a pronoun is normally placed in the more prominent position, i.e. earlier in the sentence, than a noun. For instance, if a transitive sentence has a nonpronominal NP in A function and a pronominal NP in 0 function the pronoun is favoured over the noun in initial position as is illustrated for instance in (240), (241 ), (248) and (249). Since NPs can be split up it is also possible, and often occurs, that the pronominal component of a NP stands in an early position with the rest of the NP following somewhere towards the end as in (235). The reverse occurs only in 'reprise' constructions (see §5.4.1 ).

Temporal words, including interrogative 'when', usually occur in sentence-initial position, whereas there is no preference for the placement of location words and interrogative 'where' in an unmarked construction. A location word in conjunction with a locative NP normally precedes this NP but does not have to be immediately adjacent to it, e.g. (278) and (279).

Interrogatives 'who', 'what', 'which one', 'how many', 'how' and 'what kind' fill the slot of the particular constituent that the speaker enquires about, which may be anywhere in the sentence. Interrogative 'why' always occurs in initial position. (For interrogative words see §3.6.1; for further examples of questions see §4.8.)

4.2 Clause-level derivations

In Kuku Yalanji we can distinguish two types of derivational processes, word-level derivations and clause-level derivations. As word-level derivations I consider those processes which change or add to the semantic content of a lexical item without, however, affecting the syntactic structure of a sentence, as for instance plural formation, 'personal characteristic' and 'degree' affixes, and verb reduplication. Verb concord with plural S, although not a word-level derivation in the above sense, also has no influence on syntactic structure. Such derivations have been discussed in §3.2.3 and §3.8.5.5.

A clause-level derivation, on the other hand, changes the syntactic structure in some minor or major way. Some of these derivations are optional while others are obligatory because of syntactic constraints on the surface structure of sentences. The following subsections discuss each clause-level derivation in turn, with a summary following in §4.2.5.

4.2. 1 Inchoative constructions

�-------------------------------------.------

An inchoative construction optionally verbalises the nominal predicate in nominal sentences of types (a), (b), (d), (e), (f) and (i) described in §4.1 .5 above. Inchoative verbalisers -ma-l, attached to noun, adjective or adjectival particle, and -maniji-y, attached to a noun (see §3.8.5.3) change such quasi-intransitive sentences to true intransitive sentences of the structure:

$$[NP1]S \begin{cases} noun \ adjective \ particle \end{cases} -ma-l \text{ OR } (-maniji-y)$$

meaning:

s becomes	{ noun adjective particle }
-----------	-------------------------------------

• (3 1 4) Yinya mayi ngulkurr-ma-ny. that.ABS(S) food.ABS(S) good-INCHO-PAST 'That food has become good.' (cf. (295))
• (3 1 5) Nyulu kamba-maniji-ny. 3sg.NOM(S) old.woman-INCHO-PAST 'She has become an old woman. '
• (3 1 6) Kami bayan-ba-ma-l. MMIFF.ABS(S) house-LOC-INCHO-NONPAST 'Granny is moving into the house. ' (cf. (296))
• (3 1 7) Jana karrkay bada-ma-l. 3pl.NOM(S) child.ABS(S) downwards-INCHO-NONPAST 'The children are moving down.' (e.g. to the river) (cf. (297))
• (3 1 8) Nyungu kalka dudu-ma-ny. 3sg.POSS.ABS(S) spear.ABS(S) blunt-INCHO-PAST 'His spear became blunt.' (cf. (300))
• (3 1 9) Juma juku ngakun nganka-ji-maniji-y. soon tree.ABS(S) flame.tree.ABS(S) flower-COMIT-INCHO-NONPAST 'Soon the flame tree will get flowers.' (lit. become flowerful) (cf. (303))
• (320) Nyulu kiru-kari-ma-l. 3sg.NOM(S) brain-PRIV-INCHO-NONPAST 'He is getting stupid. ' (cf. (305))
• (321) Maral yinil-ma-ny. girl.ABS(S) afraid-INCHO-PAST 'The girl became frightened. ' (cf. (3 1 2))

4.2.2 Causative constructions

There are two optional transitivising processes in Kuku Yalanji with causative effect: state causative and action causative. Both change an underlying S to O.¹⁷

State causative transitivisers -bunga-l and -kanga-l form transitive compound verbs from a nominal base or adjectival particle. (For compound formation and semantic differences between -bunga-l and kanga-l see §3.8.5.1 (i).) Subject to semantic plausibility it is possible to change an underlying nominal sentence into a transitive sentence of the structure:

¹⁷ In some Australian languages transitivising processes have the effect $S &gt; O$ or $S &gt; A$ according to the semantic content of the transitivised verb. (See Dixon 1 980:443.)

Examples:

• (322) Dingar-angka nyungu kalka dudu-bunga-ny. man-ERG:pt(A) 3sg.POSS.ABS(S) spear.ABS(O) blunt-CAUS-PAST 'The man has made his spear blunt.' (cf. (300»
• (323) Bilngkumu-ngku yinya jalbu manyarr-bunga-ny. crocodile-ERG:pt(A) that.ABS(O) woman.ABS(O) wife-CAUS-PAST 'The crocodile made that woman his wife.' (from a tale about an Annan River crocodile who abducted a woman)
• (324) Dubu-ngku maral yinil-kanga-ny. ghost-ERG:pt(A) girl.ABS(O) afraid-CAUS-PAST 'The ghost frightened the girl.' (cf. (3 1 2»

Apparently it is not possible to transitivise an inflected, e.g. locative, or a derived, e.g. comitative, nominal.

Action-causative transitiviser y-mani-l (see §3.8.5.1 (ii» forms a transitive compound verb from an intransitive verb, thus changing the structure of the underlying intransitive clause

intransitive verb

into

[NP2]A [NP1]O intransitive verb -y-mani-l

meaning: A causes 0 to verb.

Examples:

(325)	Muyar-abu	nganya	yararri-mani-l.
	wind-ERG(A)	1sg.ACC(O)	shiver-CAUS-NONPAST
	'The wind makes me shiver.'

• (326) Charlie-ngka yarraman warndi-mani-ny. Charlie-ERG:pt(A) horse.ABS(O) get.up-CAUS-PAST 'Charlie made the horse get up.'
• (327) Yinya-ngka bama-ngka wulman-angka that-ERG:pt(A) Aborigine-ERG:pt(A) old.man-ERG:pt(A)

buliman nyunba-y-mani-ny. police.ABS(O) tum. back -CA US-PAST 'That old Aboriginal man made the police tum back.' The form -y-mani-l has also been attested with two transitive verbs in the Yalanji dialect. It does not, however, have an apparent effect on the syntactic structure of the clause, but rather adds a semantic aspect. Consider:

(328) a. Ngayu	kaya	nubi-l-nubi-l.
1sg.NOM(A)	dog.ABS(O)	look.for-l-REDUP-NONPAST
'I am looking for [my] dog.'
b. Ngayu	kaya	nubi-mani-ny.
1sg.NOM(A)	dog.ABS(O)	look.for-CAUS-PAST
'I lost [my] dog.' (lit. I caused my dog to be looked for?)
(329) a. Dingar-angka	yinya	jalbu ngaki-ny.
man-ERG:pt(A)	that.ABS(O)	woman.ABS(O) hide-PAST
'The man hid the woman.'
b. Dingar-angka	yinya	jalbu ngaki-mani-ny.
man-ERG:pt(A)	that.ABS(O)	woman.ABS(O) hide-CAUS-PAST
'The man eloped with the woman.' (lit. He caused her to hide with him?)

In both a. and b. examples the A and O NPs remain in their respective functions; the only difference appears to be the added semantic aspect of some kind of 'causation' in the b. sentences.IS

4.2.3 Reciprocal

As was shown in §4. 1 .4.3, the core of a transitive sentence in Kuku Yalanji contains a NP in A function and a NP in 0 function. These NPs must not be co-referential in the surface structure. Thus if there is underlying co-referentiality between A and 0 NP as in reflexive (see §4.2.4.2) and reciprocal constructions, the surface form cannot be a transitive sentence.

The concept of reciprocity involves at least two agents which act on each other, as in: 'The dog bites the cat and the cat bites the dog', which can be represented as:

[NP1]A tr V [NP2]O [NP2]A tr V [NP1]O

Here we have two sets of co-referential NPs, NP1 and NP2. In order to give this construction an acceptable surface form NP1 and NP2 are conjoined into NP3 'dog and cat'. NP3 appears in S function and the verb is intransitivised by the reciprocal marker -wa-y (see §3.8.5.4 (i)), giving:

tr V-wa-y

The above sentence in the reciprocal Kuku Yalanji form is:

(330)	Bula	kaya	buji	bayka-wa-y.
	3du.NOM(S)	dog.ABS(S)	cat.ABS(S)	bite-RECIP-NONPAST
	'Cat and dog are biting each other.'

¹⁸ In Dyirbal the transitive verbaliser -ma-l ~ -mba-l may also attach to transitive verbs adding an 'instrumentive' aspect. But unlike the Yalanji examples, this suffix does have a syntactic effect in Dyirbal (see Dixon 1 972:95).

1 44 Chapter 4

The underlying form for a reciprocal construction may involve an unspecified number of NPs which are understood to generally act on each other, without each referent necessarily acting on each individual other referent as for instance in:

(331) Jana dingkar kuni-wa-y. 3pLNOM(S) man.ABS(S) hit-RECIP-NONPAST 'The men are fighting (lit. hitting each other).'

Thus NP3 above has to be understood to refer to all members of the set of underlying A and 0 NPs (see Dixon 1 972:2 1 3-2 1 4).

To specify in the surface form that two agents are acting on each other, Kuku Yalanji speakers use a dual pronoun, as in (330) or:

(332) Bula kaya bija-wa-y 3du.NOM(S) dog.ABS(S) lick-RECIP-NONPAST 'The two dogs are licking each other. '

A plural S NP expressed by either a pronoun or a reduplicated noun or noun + -karra (see §3 .2.3.5) implies that an unspecified number of agents is engaged in a general reciprocal action as in (3 31) or:

(333) Jana bayja-wa-y. 3pLNOM(S) ignore-RECIP-NONPAST 'They ignore each other.'

(See also examples in §3.8.5.4 (i).)

4.2. 4 Transitive verb + -ji- constructions

4.2.4. 1 Preliminary remarks

All verbs in Kuku Yalanji are strictly transitive or intransitive (see §3.8.3). However, since the verb form itself does not indicate its transitivity value, a judgement on the transitivity of the sentence has to be based on the accompanying A, 0 or S NPs. But even then it is not always obvious whether a sentence is transitive or intransitive, because: (a) pronouns show no formal distinction in transitive or intransitive subject (A or S) function while nouns are not formally distinct in intransitive subject (S) or transitive object (0) function, and (b) A or 0 may be deleted in context-related utterances. Sentences (334a) and (335a) are therefore ambiguous on first sight:

• (334) a. Nyulu bulnga-ny. 3sg.NOM shake-PAST 'He/she shook.' OR 'He/she shook something. '
• (33 5) a. Maral bulnga-ny. girLABS shake-PAST 'The girl shook.' OR 'The girl was being shaken.'

These sentences can be disambiguated by substituting pronoun and noun equivalents in the respective NPs. The correct noun form in (334) and pronoun form in (335) will be:

(334) b. Maral-angka bulnga-ny. girl-ERG:pt(A) shake-PAST 'The girl shook [something].' (33 5) b. Nyunguny bulnga-ny. 3sg.ACC(O) shake-PAST 'He/she was being shaken [by someone].'

This indicates that bulnga-l 'shake' is in fact a transitive verb in Kuku Yalanji. If, on the other hand, nyulu and maral in (334a) and (335a) could be substituted for each other without change in case marking, bulnga-l would be an intransitive verb, since the only common function of the above noun and pronoun forms is intransitive subject (S).

Under some conditions transitive sentences have to be intransitivised in Kuku Yalanji and again noun/pronoun substitution of core syntactic NPs assists in distinguishing between their function in a truncated transitive clause, like (334a) or (335a), and its intransitivised version.

There are four basic conditions for surface transitive sentences in Kuku Yalanji 19:

• (1) A NP and 0 NP must not be co-referential;
• (2) the described action must be intentional;
• (3) the agent must be stated and should be the most prominent clause constituent;
• (4) the described action must be discrete and performed on a specific object.

If any of these conditions is not met the underlying transitive sentence is intransitivised in surface structure and A and 0 NP assigned to different cases than in a transitive sentence. If transitive sentence condition (1) is not met a reciprocal or reflexive construction is required. If transitive sentence conditions (2) and (3) are not complied with the sentence is passivised. If transitive sentence condition (4) is not met an antipassive20 construction is used.

There are two processes in Kuku Yalanji by which intransitive verb stems can be derived from transitive bases. One is reciprocal derivation with -wa-y (discussed in the previous subsection) and the other is intransitive verb stem formation with -ji-²¹ which is used in the three functions of reflexive, passive and antipassive. Reflexes of a reconstructed pA form *-DHirri-y (Dixon 1980:447-449) are commonly used in reflexive, reciprocal and antipassive constructions in Australian languages, i.e. in a basic intransitivising function. (But note that Kuku Yalanji has a separate form for reciprocal which is not found in any of its related languages.) Passive constructions, on the other hand, are unusual in ergative Australian languages. Intransitive sentences with verb + -ji- forms in Kuku Yalanji, the conditions under which they are used and their syntactic and semantic implications, are discussed in tum in the following sections.

4.2.4.2 Reflexive

A transitive construction in which the A NP and O NP are co-referential, which means that an agent performs an action on himself, cannot appear as such in surface form in Kuku Yalanji. The underlying structure

[NPI JA tr V [NPI Jo

has to be intransitivised with the verbal suffix -ji-. Similar to a reciprocal construction, the A and 0 NP together assume S function in surface form, giving:

¹⁹ Similar transitivity conditions have been established for Guugu Yimidhirr (Haviland 1979a:134). Yidiny (Dixon 1 977a:276) and Djabugay (patz 1991 :297-300).

²⁰ See Silverstein (1976:140).

²¹ See also R. Hershberger (1964b:46-49). For stem formation with -ji- see §3.8.5.4 (ii).

[NPtJS tr V-ji-y

meaning: S tr V himself.

Subject to semantic plausibility a reflexive sentence can imply an intentional or unintentional action, but there are ways to clearly distinguish between these two readings. Some reflexivised verbs are inherently intentional by virtue of their semantic content, such as for instance julurri-l 'wash':

(336) Karrkay julurri-ji-y. child.ABS(S) wash-REF-NONP AST 'The child is washing itself. '

With other verbs such as 'scratch', 'hit', or 'cut' it may be ambiguous whether the reflexive action is performed on purpose or by accident. The unmarked reading, out of context, apparently tends to be 'intentional', but to make this perfectly clear one can use a reduplicated verb form indicating ongoing action, which is unlikely to happen by accident.

• (337) Bunjil yaka-n-yaka-ji-ny. widow.ABS(S) cut-n-REDUP-REF-PAST 'The widow kept cutting herself.' (It was customary for a woman to inflict a number of skin cuts on herself on the death of her husband.)
• (338) Nyulu yirrba-n-yirrba-ji-y. 3sg.NOM(S) scratch-n-REDUP-REF-NONPAST 'He keeps scratching himself. ' (perhaps because of mosquito bites)

On the other hand, the use of adverb kanban 'by mistake' unambiguously expresses the accidental nature of an action.

(339) Warru kanban kuni-ji-ny. yg.man.ABS(S) by. mistake hit-REF-PAST 'The young man hit himself by mistake.' (perhaps while hitting at something else)

Usually, contextual information helps to distinguish between an intentional action and a misfortune, as has been suggested in the translations of (338) and (3 39).

(340)	Bunjil	yaka-ji-ny	dunyu-nku.
	widow.ABS(S)	cut-REF-PAST	husband-DAT:pt
	'The widow cut herself because of her husband.' (i.e. his death)
(341)	Nyulu	wulbuman	jina
	3sg.NOM(S)	old.woman.ABS(S)	foot.ABS(S)
	mayi	baka-l-baka-nya.	baka-ji-ny
	plant.food.ABS(O)	dig-l-REDUP-SUB22	poke-REF-PAST
	'The old woman poked her foot while digging for food.' (e.g. yam)

Sentence (341) also illustrates the use of reflexive where a specified body part is affected. Another example is:

²²For subordinate constructions of this type see §4.4.4. 1.

(342) Ngayu yimbi junka-ji-y. 1 sg.NOM(S) lips.ABS(S) paint -REF-NONP AST 'I paint my lips.' (i.e. put lipstick on)

(For inalienable possessive constructions see §4.S.2).

As I have shown before when discussing the instrumental case (§4. 1 .4.4 (i)), an inanimate agent does not change its case marking, i.e. instrumental, when the sentence is passivised as in (248b) and (249b) and

(343) Nyulu kaja-ji-ny baduru-bu. 3sg.NOM(S) tie-ITR-PAST fishing.line-INsT 'He got tangled up in the fishing line. '23

But consider the similar reflexive construction:

(344)	Nyulu	jalbu	naybu-bu	yaka-ji-ny,
	3sg.NOM(S)	woman.ABS(S)	knife-INST	cut-REF-PAST
	minya	yaka-l-yaka-nya.
		meat-ABS(O) cut-l-REDUP-SUB
		'The woman cut herself with a knife while cutting meat.'

Because of the parallelism in structure it would be possible to interpret (248b), (249b) and (343) as reflexive rather than passive constructions. This would also be plausible from a semantic point of view in that where an inanimate instrument is involved in an action on an animate patient, it is in fact the animate patient who can often be held responsible for the action. In (344) the woman as S NP of a reflexive sentence is both agent and patient: she cuts herself by being careless. In the same way, the S NPs in (248b), (249b) and (343) could be seen to indirectly inflict the misfortune of getting hit, burnt or tangled up upon themselves by being inattentive or clumsy. This interpretation would, correctly, deny the inanimate instrument/agent any volitional control over the action. (A similar concept is expressed in colloquial English reflexive-cum-causative construction which would be an appropriate translation for the above sentences, that is 'he got himself burnt by a stinging tree' or 'he got himself tangled up in the fishing line'.) I regard reflexive as a possible alternative interpretation of sentences like (248b), (249b), and (343), but the evidence for this is purely semantic; syntactically they could be agentless passive just as well as reflexive.

4.2.4.3 Passive

A passive construction characteristically shifts the focus of attention from the agent, i.e. the transitive subject (A), of a transitive sentence to the patient (0) and hislher state of being acted upon (see Lyons 1 968:372). Whatever mechanisms are employed in different languages to achieve this shift of focus, the resulting 'passive' sentence is a superficially intransitive sentence in which the underlying patient (0) is the intransitive subject (S) and thus usually the discourse topic. The underlying agent, being reduced to an ancillary role, appears in a peripheral syntactic case, if indeed it is mentioned at all. True to the 'patient focus' nature of passives, expression of the agent is not obligatory in any language (see Silverstein 1 976: 140). Nor are there syntactic reasons to include the underlying agent in the surface

²³ For passive formation see next subsection.

148 Chapter 4

structure, since the passive as a derived intransitive construction only requires the core elements of intransitive subject (S) and an intransitive verb, all other constituents being optional.

�--------------------.------------------------------------------------------------------�

A passive construction in Kuku Yalanji requires intransitivisation of the transitive verb by stem-forming suffix -ji-. The underlying O appears in absolutive case if a noun, and nominative case if a pronoun, and the underlying A receives locative case marking in surface structure. Thus from an underlying sentence

tr Y

we obtain

$$\begin{array}{c} \text{[NP2] } \text{ABS ls]} \ \text{[NOM]} \end{array} \quad \text{tr} \begin{array}{c} \text{tr} \begin{pmatrix} \text{V} \cdot \dot{j} \text{\_} \text{[NP1]} \text{LOC} \end{pmatrix} \end{array}$$

which means: 's is being V by NPI ' (NPI being optional). A transitive sentence and its passive version are illustrated below:

• (345) a. Yaburr-ungku warru bayka-ny. shark-ERG:pt(A) yg.man.ABS(O) bite-PAST 'A shark bit the young man.'
  • b. Warru (yaburr-undu) bayka-ji-ny. yg.man.ABS(S) shark-Loc:pt bite-ITR-PAST 'The young man was bitten (by a shark). '

The passive in Kuku Yalanji does not entirely fit the traditional definition of a patienttopicalisation process as given above. It may promote the patient to focus of attention, but it does not necessarily do so. The primary means of topicalisation in any sentence type in Kuku Yalanji is to place the topic in leftmost position. Topic choice and the significance of word order are discussed in §5.3.2.1 and §5.4.1.

One can say that the passive in Kuku Yalanji is semantically motivated or discourseoriented according to which transitive sentence condition is the prompting factor. But it is never necessary to use the passive because of syntactic co-referentiality restrictions in sentence co-ordination, although of course a passive can occur in co-ordinated sentences. According to their prompting condition, there are two types of passive in Kuku Yalanji, which may be termed 'chance passive' and 'topic passive'.

The chance passive is employed where transitive sentence condition (2) (see §4.2.4.1) is not met, that is where a potentially controlling agent performs an action unintentionally. Here the underlying agent is always retained in surface structure, albeit in locative case. In the unmarked word order the agent in locative case appears in leftmost position which means that it does not lose its discourse prominence. Note also that translations of chance passive sentences given by informants did not show passive in English. The involuntary aspect was always rendered as 'happened to'.

(346)	Jalbu-ndu	jarba	baka-ji-ny.
	woman-LOC:pt	snake.ABS(S)	poke-ITR-PAST
	'The woman happened to poke a snake.'

(347) Bulki-nda ngunJa-Jl-ny mayi ngayku cattle-LOc:pt forage-ITR-PAST plant.food.ABS(S) l sg.POSS.ABS(S) mayngku. mango.ABS(S)

'The cattle happened to forage my mangoes. '

• (348) N gayku-ndu yundu nyajl-Jl-ny. I sg.LOC:pt 2sg.NOM(S) see-ITR-PAST 'I happened to see you.'
• (349) Dingkar-anda ngayu jina narri-ji-ny. man-LOC:pt I sg.NOM(S) fOOt.ABS(S) step.on-ITR-PAST 'The man happened to step on my foot.'
• (350) Babi-nda wukay dirka-bunga-ji-ny. FM-LOC:pt yam.ABS(S) fine-CAUS-ITR-PAST 'Grandmother happened to smash the yam to tiny pieces.' (e.g. by digging carelessly)

The chance passive has only been attested with past tense which is plausible since an accidental action can only be reported in retrospect.

The important characteristic of the chance passive is that the expressed action as such is unpremeditated and unintentional, in other words that 'it just happens'. In (346), for instance, the woman may have been sweeping the yard and in doing so poked a hidden snake with the broom. In (347) the cattle were just wandering about the place at night and chanced on the mangoes under my caravan and proceeded to rout them out. A context for (348) could be that the speaker was walking along minding his own business and happened to see the addressee driving past in a car. This is in contrast to 'accidental' action as expressed in a transitive sentence with adverb kanban 'by mistake'. In this event the action itself is intentional, but it has an unintentional effect. For instance, compare the passive construction in (349) with the transitive sentence using kanban below:

(35 1) Dingkar-angka nganya jina kanban narn-ny. man-ERG:pt(A) I sg.ACC(O) foot.ABS(O) by. mistake step.on-PAST 'The man accidentally stepped on my foot.'

This would imply that the man was intentionally stepping or stamping on something, perhaps ants on the ground, and in doing so accidentally caught my foot. Or a transitive sentence with kanban for (346), that is

(352) lalbu-ngku jarba kanban baka-ny. woman-ERG:pt(A) snake.ABS(O) by. mistake dig/poke-PAsT 'The woman accidentally poked a snake.'

could imply that the woman was digging for scrub-hen eggs and in doing so poked her hand into a snake. Note that speakers do not include kanban in chance passive sentences. While such constructions are not regarded as "wrong", they are thought to be odd.

The 'topic passive' may be used where transitive sentence condition (3) (see §4.2.4. 1) is not fulfilled. This typically occurs where the agent is a lower animate referent acting on a higher animate patient, or where the agent is either unknown or considered irrelevant. In the latter event the agent is omitted from surface structure. Note that topic passive sentences were rendered by a passive construction in English by informants. Examples of passive sentences with a lower animate agent acting on a higher animate patient are:

• (353) Warru wara-nda mani-ji-ny. yg.man.ABS(S) groper-LOC:pt take-ITR-PASf 'The young man was taken by a groper.'
• (354) N gayu yinil-kanga-ji-ny bilngkumu-ndu. I sg.NOM(S) fear-CAUS-ITR-PAST crocodile-Loc:pt 'I was given a fright by the crocodile.'

However, because of its free word order, Kuku Yalanji does not necessarily have to use a passive in order to topicalise the patient. It would be equally acceptable, and is often done, to simply move the 0 NP into sentence-initial position, for example:

1

(355) Warru wara-ngka mani-ny. yg.man.ABS(S) groper-ERG:pt(A) take-PASf 'The young man was taken by a groper.' (cf. (353))

or

(356) Nganya bayka-ny biringal-da. I sg.ACC(O) bite-PASf scorpion-ERG(A) 'I was bitten (stung) by a scorpion. '

On the other hand, passive can be used to topicalise an 0 NP that is at the same hierarchy level as the agent, for example:

(357) Yinya bama buliman-anda that.ABS(S) Aborigine.ABS(S) police-LOC:pt grab-ITR-PAST 'That Aborigine was grabbed by the police.' karrba-ji-ny.

Some speakers tend to prefer passive, others favour the use of transitive sentences with leftmost 0 NP, and apparently there are no speakers' intuitions on a semantic difference between a topic-passive and a topicalised transitive sentence such as (355) or (3 56). There is no choice, however, where the agent is omitted because it is unknown or irrelevant and is not recoverable from context. Here the topic-passive must be used, for example:

(358) Minya dinda-n-dinda-ji-y baya-nga. meat.ABS(S) roast-n-REDUP-ITR-NONPASf fire-LOC 'Meat is roasting in the fire.'

(Note that fire is not the agent; such an interpretation was regarded as quite ridiculous by speakers.)

(359) Store nanda-ji-y nyiku. store.ABS(S) close-ITR-NONPASf today 'The store is closed today.'

The following sentences (360) and (36 1) may also be regarded as agentless passives since the action described by the verb may be interpreted as being performed by an unspecified agent, i.e. 'anyone'.

(360) Nyulu maral ngulkurr nyaji-ji-y. 3sg.NOM(S) girl.ABS(S) good see-ITR-NONPASf 'The girl looks nice.' (lit. is good to look at [by anyone])

(361) Yanya mayi ngulkurr nuka-ji-y. this.ABs(s) food.ABS(S) good eat-ITR-NONPASf 'This food tastes nice.' (lit. is good to eat by anyone)

An agentless passive may of course also imply a chance action as in (362) below which was saId to me by an old lady with a sweet tooth when she saw me setting out with a walkabout party:

(362) Mayngku nyaji-ji-y, mana! mango.ABS(S) see-ITR -NONP ASf take-IMP ''[If] [you or someone] happen to see any mangoes, take [bring] them!'

R. Hershberger ( 1 1964b:47, 1979:59) reports that the verbs nuka-l ‘eat', baka-l 'dig, poke', yajarri-l 'laugh about' and muja-l 'gather' do not occur in passive, but only in antipassive constructions (see below). I have found that this is not so. Examples of nuka-l and baka-l in passive are given in (36 1) and (346) above. There are, however, semantic restrictions on the use of these verbs in a chance passive because they refer to inherently volitional actions and are therefore unlikely to happen by chance (except baka-l in the sense of 'poke'). If the wrong object were accidentally affected, a transitive sentence with kanban would be used rather than the passive (see above). But these verbs may occur in a topic passive, such as nuka-l 'eat' in (36 1) or munja-l 'gather' in the next sentence.

(363) Marra munja-ji-ny zamia.nut.ABS(S) gather-ITR-PAST 'Zamia nuts were gathered [by someone].'

With yajarri-l 'laugh about', on the other hand, a passive, or anti-passive for that matter, is ambiguous. Consider:

(364)	Nyulu	maral	yajarri-ji-ny	warru-ndu.
	3sg.NOM(S)	girl.ABS(S)	laugh.about-ITR-PAST	yg.man-LOC:pt

This could be passive,

'The girl was ridiculed (lit. laughed about) by the young man.'

where nyulu maral has underlying O- and warru A-function. But it could just as well be 'antipassive'

'The girl laughed at the young man [in a friendly way]',

where nyulu maral has underlying A and warru O function. Speakers favour the latter interpretation, which coincides with R. Hershberger's theory at least for this verb. But it is not at all certain that this construction is indeed a true antipassive, hence the inverted commas. This problem will be discussed further in the next subsection.

4.2.4.4 Antipassive

The so-called antipassive is regarded as the ergative language's answer to the passive of the accusative language. According to Silverstein (1976: 140)

... in accusative systems, the patient is regularly expressed in direct transitive constructions by the 'unique' case marking ... the accusative, and in passive constructions this assumes the 'paired' case form, the nominative. Ergative systems have an analogous construction, here termed antipassive , which has all the

1 52 Chapter 4

properties of the passive ... The 'unique' case here is the ergative, coding the unique function of direct transitive agent (A), and in antipassive forms the transitive agent is expressed by a surface absolutive (or nominative) case marking, the verb has a change in voice, with a special mark, the transitive object (normally coded by surface absolutive case) appearing at most facultatively in some oblique, adverbial case marking.

Kuku Yalanji does have a construction with the formal properties of an antipassive, but its semantic and pragmatic properties are quite different from that of the passive. It is remarkable that Kuku Yalanji uses both passive and antipassive constructions, a characteristic which is rare but which it shares with its northern neighbour Guugu Yimidhirr.24

In an antipassive construction, as in passive, the transitive verb is intransitivised by stemforming suffix -ji-. The underlying agent (A) appears in S function, taking absolutive or nominative case marking for a noun or pronoun respectively. The underlying transitive object (0) takes locative case in surface structure or even, where appropriate, perlative case. Thus from an underlying transitive structure

	[NP1]A	tr V	[NP2]O
we obtain	[NP1 ABS] S	tr V-ji-	[NP2 LOC]
	NOM		PERL

The semantic content of an antipassive construction cannot be captured in one formula, but ample exemplification will be given below. An example of the structural difference between underlying and surface form is:

(365) a. Nyulu	dingkar-angka	minya	nuka-ny.
3sg.NOM(A)	man-ERG:pt(A)	meat.ABS(O)	eat-PAST
'The man ate meat.'

b. Nyulu dingkar minya-nga nuka-ji-ny. 3sg.NOM(S) man.ABS(S) meat-LOC eat-ITR-PAST "The man had a good feed of meat (he wasted nothing)."

In syntactically ergative languages like Dyirbal and Yidiny the antipassive is obligatory for sentence co-ordination with co-referential S and A; that is, the sentence containing the A must be antipassivised so that in surface structure co-referential NPs are both in S function (Dixon 1 972:73-74, 1 977:278-279). This is not so in Kuku Yalanji where, as far as coordination is concerned, almost anything goes (see §4.3.1 ). Like passive, antipassive may of course occur in co-ordinated structures, but this is not of syntactic relevance; the antipassive has a different meaning from the transitive sentence.

The antipassive in Kuku Yalanji is used productively for what may be called a 'generalised action', a term coined by Haviland (1 979a:1 32) for the similar use of antipassive in Guugu Yimidhirr. This means that the described action is not discrete and is performed on some general or 'non-individuated'25 object which may or may not be stated in surface structure, that is where transitive sentence condition (4) is not met.26 To indicate

26 Djabugay uses the cognate intransitive suffix -yi- under a similar condition, among other conditions, without however applying antipassive (see Patz 1 99 1 :299).

²⁴ The functions of passive and antipassive in Guugu Yimidhirr are similar to those in Kuku Yalanji (see Haviland 1 979a: 1 28- 1 34).

²⁵ See Hopper and Thompson ( 1 980:253ff.); this term is discussed further later on in this subsection.

'generalised action' on a 'non-individuated' object the antipassive construction contains a reduplicated verb form as for instance in text 3, lines 39 and 48 and:

• (366) Yinya karrkay kaya-nda kuni-n-kuni-ji-y. that.ABS(S) child.ABS(S) dog-LOC:pt hit-n-REDUP-11R-NONPAST "That little one is hitting all the dogs (around here)."
• (367) Barna dunga-ny bunjurri-ji-ny. Aborigine.ABS(S) go-PAST throw .spitJcurse-11R -PAST 'The Aborigine went and threw curses everywhere. '
• (368) lalbu wukay-rnba nubi-nubi-ji-y. woman.ABS(S) yam-LOC search.for-REDUP-11R-NONPAST 'The woman is looking around for yam.'

(Note that nubi-ji-y in a reflexive construction means 'be lost'.)

(369) lalbu bayan-ba yindu-yinduy-rnbu nuri-nuri-ji-y. woman.ABS(S) house-LOC other-REDUP-LOC peep-REDUP-1TR-NONPAST 'The woman is having a sticky-beak in all the other houses.'

For (366) the generality of the action was made explicit by my informant, whose translation is given above; a similar situation obtains for (369) where the actual houses that are investigated by the woman are not identified. Sentence (368) may be contrasted with

(370) lalbu-ngku wukay nubi-l. woman-ERG:pt(A) yam.ABS(O) search.for-NONPAST 'The woman is searching for yam.'

where the action is more specific and limited than the casual but far-reaching 'looking around' in (368). In (367) the unfortunate targets of the raving man are not even mentioned. Note that as far as co-ordination is concerned, (367) would be equally acceptable as

(37 1) Barna dunga-ny bunjurri-ny. 'The Aborigine went and threw a curse.'

The difference is that in (371) the man cursed only once.

In the Nyungkul dialect, which makes a distinction between locative and perlative, the perlative may also be used in an antipassive construction to express the concept of 'through'. For instance:

(372) Kaya rnayi-burr ngunja-n-ngunja-ji-y. dog.ABS(S) food-PERL forage-n-REDUP-1TR-NONPAST 'The dog is foraging through the food.'

and instead of locative in (369):

(373) lalbu bayan-darr yindu-yinduy-rnburr nuri-nuri-ji-y. 'The woman is having a look through all the other houses.'

If a simple (i.e. not reduplicated) verb form is used in an antipassive in Kuku Yalanji this expresses the semantic aspect of 'total effect on the patient'.27 One such example is (365b)

²⁷ This is in contrast to the interpretation of antipassive as having "partitive" meaning (see Hopper and Thompson 1 980:263). For further comments see §4.2.5.

1 54 Chapter 4

where the man is considered to have "wasted nothing" in the words of my informant. Consider also:

• (374) Ngamu mayi-nga wambi-ji-ny. mother.ABS(S) food-LOC share.out-ITR-PAST 'Mother shared out all the food.'
• (375) Nyulu jalbu ngayku-wun-bu bayan-ba ngunja-ji-ny. 3sg.NOM(S) woman.ABS(S) I sg.POSS-LOC house-LOC take.over-ITR-PAST 'The woman took over my house lock stock and barrel.'

Ngunja-l, also glossed as 'forage' in (347) and (372) refers to the action of claiming or obtaining something that is, or is thought to be, discarded or nobody's particular property.

Use of antipassive with the semantic implications exemplified above is quite productive in Kuku Yalanji. Furthermore, some broadly termed 'verbs of communication' may be intransitivised and in this form require a syntactic construction that is identical to antipassive, i.e. the agent appears in S function, underlying 0 in locative case and the goal of the verbal interaction is added in dative case. The semantic difference between the transitive and the intransitivised verb form may again be interpreted as 'direct' versus 'generalised' action, i.e. the underlying patient is not solely and directly affected by the action but is demoted to an ancillary role. Because the semantic link may be regarded as somewhat remote, particularly in the case of baba-l/babaji-y (see below), I regard the forms in the second column below as 'fixed' intransitive stems similar to the fixed reflexive stems listed in §3.8.5.4 (ii). Consider:

kunja-l	'call, summon'	kunjaji-y	'ask someone for something'
walngkurri-l	'bark at (tr)'	walngkurriji-y	'pester someone for something'
baba-l	'try, taste'	babaji-y	'ask someone about something'

Janjarri-l 'laugh about, ridicule' andjanjarriji-y 'laugh at in a friendly way' can probably also be included in this list. The link between baba-l and babaji-y can perhaps be interpreted as: babaji-y '� someone's knowledge about something'. The following examples illustrate ^I the use of these intransitive stems:

• (376) a. Dingkar-angka nganya kunja-ny. man-ERG:pt(A) I sg.ACC(O) call-PAST 'The man called me.'
  • b. Dingkar ngayku-ndu mayi-ka kunjaji-ny. man.ABS(S) I sg-LOC:pt food-DAT askJor-PAST 'The man asked me for food. '
• (377) Karrkay ngamu-ndu walngkurriji-ny child.ABS(S) mother -LOC:pt pester .for-PAST 'The child pestered mother for a lolly.' lolly-ka. lolly-DAT
• (378) Buhman warru-ndu babaji-ny kamukamu-ku. police.ABS(S) yg.man-LOc:pt ask.about-PAST alcohol-DAT 'The police asked the young man about the grog.'

In a synchronic study I hesitate to interpret these intransitive sentences as antipassive derivations of underlying transitive constructions. But I would suggest the possibility that at an earlier stage this may have been so and that subsequently the intransitivised verb form has shifted further from the semantic content of the transitive form, being stuck, as it were, with the apparently antipassive construction.

4.2.5 Summary and conclusions

To summarise, Kuku Yalanji has an inventory of six derivational processes at clause level, which can be distinguished as optional and obligatory derivations. An optional derivation primarily aims at changing the semantic content of a sentence; a change in syntax is incidental to this primary aim. Optional derivations are inchoative, causative and antipassive. Obligatory derivations are syntactically motivated, mainly by restrictions on the co-referentiality of A and 0 NP in surface structure. Reciprocal and reflexive are derivations of this type.

Passive has a triple function as optional 'chance passive' and 'topic passive' which in tum may be optional or obligatory in certain circumstances. Topic passive can be used optionally to topicalise the 0 NP in a surface structure containing both A and 0 NP. (A transitive sentence with leftmost 0 NP is apparently equally acceptable for this purpose.) Obligatory use of topic passive applies where the underlying agent is omitted in surface structure and this agent is not recoverable from context.

Passive and antipassive with a reduplicated verb form indicate a lower degree of transitivity, i.e. unintentional action, action performed by an unknown or irrelevant or lower animate agent, and generalised action. All this is in accordance with the Hopper and Thompson (1 980) transitivity theory. However, antipassive in Kuku Yalanji does not necessarily have "partitive" meaning, which Hopper and Thompson regard as a characteristic feature for antipassive. "Partitive" may apply for the aspect of generalised action as expressed by the reduplicated verb forms in (366)-(369). However, in sentences (365b), (374) and (375) antipassive with a simple verb form expresses rather the opposite, namely 'total effect'. As in generalised action antipassive, the patient in total-effect antipassive constructions is 'non-individuated', e.g. generic terms in (365b) and (374), and 'my house' (not every bit of furniture specified) in (375). 'Non-individuated' object is thought to be another indication of low transitivity (Hopper and Thompson 1 980:253ff.), but it is not a characteristic feature of antipassive in Kuku Yalanji since it may apply just as well in transitive sentences. Thus antipassive in Kuku Yalanji has the unusual dichotomy that it implies higher transitivity (total effect) or lower transitivity (generalised action) than a corresponding transitive sentence.

Since passive and antipassive in Kuku Yalanji have almost entirely semantic motivation and are not required for syntactic purposes in subordination and co-ordination their existence is no indication of whether Kuku Yalanji has ergative or accusative syntax. To anticipate the findings in the next subsections and also in §5.3.2. 1 -§5.3.2.4 there is no discernable pivot NP in Kuku Yalanji, such as AlS in accusative Western Australian languages (Dench 1 982) or S/o in ergative Dyirbal (Dixon 1 979:62-63), according to which coordinated and subordinated clauses have to be modified. This freedom in co-ordination and subordination could be an indication that the language is syntactically in flux, containing aspects of both ergative and nominative/accusative syntax.28

²⁸ Similar situations are found in Guugu Yimidhirr (Haviland 1 979a:1 28ff., 1 56ff.) and Yidiny, about which Comrie (1981: 1 08) comments: " ... the Yidiny material illustrates particularly clearly ... that it is misleading to classify a language as being either ergative or not, rather one must ask: to what extent, and in what particular constructions is the language ergative ... " .

4.3 Sentence co-ordination

Co-ordinate structures in Kuku Yalanji consist of two or more clauses of independent status, i.e. clauses which may just as well occur on their own, that are linked together by a specific intonation pattern. The distinguishing feature of co-ordinate versus subordinate constructions in Kuku Yalanji is the verbal inflection. In co-ordinate sentences we find only past or nonpast tense marking, irrealis, or imperative. All other verbal inflections mark a clause as subordinate (see §4.4). (While purposive may occur in an independent clause, two co-ordinated clauses cannot both have purposive inflection.)

Usually, in a co-ordinate structure the constituent clauses are simply juxtaposed. However, there are a few particles and clitics which appear in co-ordinate sentences and add a certain semantic link to the conjoined clauses.

4.3. 1 Simple conjoining

Clauses are conjoined in discourse in order to convey some temporal or logical connection; they may or may not share a common NP.29 In co-ordinate structures without a common NP the described actions may be simultaneous or successive, whichever is the more logical interpretation. The only indication for co-ordination, as opposed to a sequence of several independent clauses, is intonation: there is a rise at the end of each non-final clause and falling intonation after the final clause. (For intonation patterns see §2.6.3.)

• (379) Ngayu jilba-dunga-ny, nyulu bayan-ba bundanda-ny. I sg.NOM(S) walk-go-PAsT 3sg.NOM(S) house-LOC sit.REDUP-PAST 'I went for a walk andlwhile he stayed at home.'
• (380) Babarr-angka mayi waju-l, jinkurr-ungku eZ-ERG:pt(A) food.ABS(O) cook-NONPAST yZ-ERG:pt(A) ngawa nyaji-l-nyaji-l. baby.ABS(O) look-l-REDUP-NONPAST 'Elder sister is cooking food andlwhile younger sister is watching the baby.'
• (38 1) N gayu yarraman karrba-l, yundu daka-l. I sg.NOM(A) horse.ABS(O) hold-NONPAST 2sg.NOM(S) climb-NONPAST 'I hold the horse andlwhile you climb up. ' (i.e. on the horse; daka-l is intransitive)

Co-ordinate structures without co-referential NPs are not all that frequently used in Kuku Yalanji. A much more common stylistic device in discourse, particularly in narratives, is the so-called topic chain. Where the S or A NP, i.e. the topic or pivot of a clause, refers to the same actor as the S or A NP of a number of immediately following clauses, it is normal practice to delete the S or A NP in all but the first clause. Logically such a topic chain refers to successive actions since it is unlikely that someone would perform several actions at the same time. Note that in Kuku Yalanji there is no restriction on the co-referentiality of A and S which means that transitive and intransitive clauses can be co-ordinated without any obligatory change in the transitivity value of the verb. (Anaphoric topic deletion in discourse is considered more fully in §5.3.2 with references to the appended texts.) Some examples of co-ordination possibilities are:

29 See 'Sequence Sentence' and 'Apposition Sentence' in Kuku Yalanji, R. Hershberger (1970:8 1 3-816).

(382) Nyulu yindu nyunba-ny-baja, kangkal-kangkal 3sg.NOM(S) other.ABS(S) turn.back-PAsT-again own.child-REDUP.ABS(O)

nyungu wubul-ku mani-ny, wundi-ny badabada 3sg.POSS.ABS(O) many.ABS(O)-EMPHI get-PAST bring-PAST down

bayan-ba, camp-LOC milbi-ny-da. shoW-PAST-TEMPI

'The other one turned back again, fetched her many children, brought them down to the camp and showed them.' (from a story about how the cassowary tricked the brolga into killing her young ones)

(383) Yundu kada-nyaku, nganya nyaji-nyaku. 2sg.NOM(S) come-IRR I sg.ACC(O) see-IRR '[If] you had come, you would have seen me.'

In the above examples the initial clause is intransitive followed by transitive ones.

(384) Bula warru-ngku nambarr kulba-ny, 3du.NOM(A) yg.man-ERG:pt(A) bamboo.ABS(O) pull. out-PAST kaja-ny, dunga-ny-baja. wrap.up-PAST gO-PAST-again 'The two young men pulled out the bamboo, wrapped it up and went on again.' (from a story of how the native bamboo came to grow all over the Cape York Peninsula)

Here the first two clauses are transitive and the last one is intransitive.

(385) Nyulu dingkar bunda-ny, minya-nga nuka-ji-ny, 3sg.NOM(S) man.ABS(S) sit-PAST meat-LOC eat-ITR-PAST wunana-ny-da. lie.down.REDUP-PAST -TEMPI 'The man sat down, had a good feed of meat and then had a rest.'

Here all three constituent clauses are intransitive. Note that the second clause is antipassive, not because the common NP has to agree in function, but to indicate 'total effect' of the action (see §4.2.4.4).

The intonation pattern of these topic chains is such that each non-final verb has either level or rising intonation and the last verb falling intonation. Theoretically, any number of actions can be joined together in this fashion as long as the topic remains the same. However, presumably to break the monotony of such a recital, sentence-final intonation is often applied not later than the third verb. If the topic still remains the same for any further actions, the new sentence may be introduced with the appropriate S/A pronoun. But an anaphoric pronoun is not obligatory, even if the intonation pattern clearly distinguishes several separate sentences, i.e. there can be zero-anaphora for topic across the boundaries of consecutive sentences.

If a number of actions are performed by the same agent on the same object, the 0 NP may also be deleted once it has been introduced as in (382) and (384). But some speakers seem to prefer to include an anaphoric pronoun in 0 function every now and again, particularly if the object is animate.

1 58 Chapter 4

Sentences may also be intonationally co-ordinated if the S NP, but apparently not the A NP, is co-referential with the 0 NP of the following sentence. It seems that zero anaphora is not possible in this event, but pronoun reduction is required.

(386) Dingkar wala-ny, jalbu-ngku nyunguny kuni-ny. man.ABS(S) enter-PAST woman-ERG:pt(A) 3sg.ACC(O) hit-PAST 'The man came in and the woman hit him.'

If, on the other hand, the second part of a sentence like (386) should be passive in order to imply a chance event, the first S NP is co-referential with the surface S NP of the following passive clause and may therefore be deleted in the second clause:

(387) Dinkar wala-ny, kuni-ji-ny jalbu-ndu. man.ABS(S) enter-PAST hit-1m-PAST woman-LOc:pt 'The man entered and the woman happened to hit [him].'

4.3.2 Co-ordination particles

Kuku Yalanji has four co-ordination particles, which are used to indicate a relationship of 'contrast', 'condition', or 'reason' between two clauses. These particles are always placed at the beginning of a conjoined clause. They are:

kari and yamba	'but, however'
kaki	'if, when'
kunka	'in case'

Another particle, nganganjirrka, may be used in free vanatIon with kunka, but more frequently occurs just by itself to express 'x does Y just in case (something might happen)'.

Clauses containing these particles cannot occur independently, at least not out of context. But it would not be justified to class them as subordinate rather than co-ordinate, because their syntactic structure is that of an independent clause. They may be termed 'dependent sentences' according to Waterhouse (1 963). Such constructions do not normally have more than two constituent clauses, which may or may not have a common A or S NP. Deletion of the second S/A NP is not favoured, although possible; pronoun reduction is preferred.

The particles kari and yamba (sometimes yambada or yambala) imply, in broadest terms, some kind of contrast between the actions or states referred to in two clauses, similar to 'but' or 'however' in English. Both particles are glossed as 'but' in translations. Although kari and yamba are freely interchangeable according to my informants, spontaneous use in elicitation and evidence from texts suggest a subtle semantic difference between them. Kari, which is also a negation particle (see §4.9.2), is typically used to introduce an event which has some negative influence on the event described in the related clause.30 The set patterns for such a co-ordinate construction is that the first clause refers to an event that is desired or under way, and the second clause, beginning with kari, contains the event which has a hindering effect on the first:

'x does/wants to do something, but something else happens', for example:

30 See R. Hershberger's (1 970:816-8 1 8) term "implied-action-diverted" for kari and yamba clauses.

• (388) Ngayu buwun-ji dunga-ny, kari muyar junkuji. I sg.NOM(S) boat-COMIT gO-PAST but wind strong 'I went out with a boat, but there was a strong wind.' (I had to tum back)
• (389) lalbu yinya wawu baduri-ji dunga-nka, woman.ABS(S) that.ABS(S) want fishing.line-coMIT go-PURP kari nyulu bambay. but 3sg.NOM(S) sick 'That woman would like to go fishing, but she is sick.' (she can't go)
• (390) N gayu nyaji-nka yinya bama, kari kalakalbay-bajaku. I sg.NOM(A) see-PURP that.ABS(O) person.ABS(O) but far.away-very 'I am trying to see that person, but [he isll am] too far away.' (I cannot see him properly)
• (39 1) Nyulu warru-ngku kuyu mani-ny, kari burrki-n-burrki-n. 3sg.NOM(A) yg.man-ERG:pt(A) fish.ABS(O) get-PAST but fidget-n-REDUP-PAST 'The young man got the fish [on the line], but [it] kept on fidgeting/struggling.' (presumably it got away)

Kari may also introduce an independent sentence in discourse to indicate a contrast to something previously stated. For more on this use of kari as 'collateral' marker in discourse see §5.5.3.

Yamba, on the other hand, is used more spontaneously than kari in the conveyance of some additional and contrastive information on a state rather than an event. Again, the yamba clause is the second constituent in a co-ordinate structure.

(392) Yinya budmandu karrkay, yamba that.ABS(S) portmanteau.ABs(s) small but 'That port (suitcase) is small, but [it] is heavy.' kulngkul. heavy

Speakers are quite happy to exchange yamba for kari and vice versa in the above sentences, so it appears that a semantic distinction, if there ever was one, is losing its significance.

The co-ordination particle kaki introduces a conditional clause and will be glossed as 'if'. Usually, the kaki clause is the first of two co-ordinated clauses, but it may also occur in second position. Co-ordinate structures with kaki apparently cannot occur in the past tense.31

(393)	Kaki yundu	yinya	kaya	juku-bu	kuni-l,
	if 2sg.NOM(A)	that.ABS(O)	dog.ABS(O)	stick-INST	hit-NONPAST
	yununy bayka-l.
	2sg.ACC(O)	bite-NONPAST
	'If you hit that dog with a stick, [it] will bite you.'
(394)	Kaki ngayu	ngiki-ji,	ngayu	yangka
	if 1sg.NOM(S)	cough-COMIT	1sg.NOM(A)	green.ant.ABS(O)
	dumbarri-l,	nuka-l.
	break-NONPAST	eat-NONPAST
	'If I have a cough/cold, I break up green ants and eat [them].'

³ 1 See R. Hershberger (1 979:75-76) 'Temporal/conditional Relator-axis Clause'.

1 60 Chapter 4

(395) Nyulu warngku-wuna-y, 3sg.NOM(S) sleep-lie-NONP AST 'He has a sleep, if he feels tired.' kaki nyulu bangkarr bajaburray. if 3sg.NOM(S) body.ABS(S) tired

To express such a conditional/temporal relationship in past tense the two clauses are simply juxtaposed, for example:

(396) Nyulu wawula-ny, nyulu warngku-wuna-ny. 3sg.NOM(S) be.exhausted-PAST 3sg.NOM(S) sleep-lie-PAST '(If/when) he was exhausted, he had a sleep.'

Particles kunka or nganganjirrka 'in case' may be used to introduce a 'hypothetical reason' subordinate clause (see §4.4.4.4), or may link a dependent sentence to a preceding one, for example:

(397) Ngayu bana wundi-l, kunka / nganganjirrka Isg.NOM(A) water.ABS(O) take-NONPAST in.case bana kari baral-darr. water.ABS(S) NEG road-PERL 'I take some water in case there is no water along the road. '

Note that if the speaker knows for sure that there is no water on the road a co-ordinated construction such as (397) is not possible, but two independent sentences are used:

(398) Ngayu bana wundi-l. Yinya-ngkarr Isg.NOM(A) water.ABS(O) take-NONPAST that-PERL baral-darr bana yambakari. road-PERL water.ABS(S) nothing 'I take some water. There is no water at all along the road. '

4.3.3 Clitics in co-ordinate constructions

Temporal clitics -da 'then', -(ng) Vrr 'first of all', and kuda 'meanwhile' (see §3.10.1) occur frequently in co-ordinated sentences to specify a particular temporal connection. However, they are not actual indicators of co-ordination. They are used just as often in independent clauses to provide a temporal reference within a wider time setting. Their discourse function in temporal cohesion is explained and exemplified in §5.2.2.

Clitic -ji ~ -yijin 'in turn' (see §3. 1 0.3), although it may also occur in independent clauses, is more relevant for sentence co-ordination. In this function it indicates an (appropriate) action that may have been prompted by the action of a different actor as stated in the coordinated clause, for example:

(399)	Nyulu 3sg.NOM(A)	ngayku 1sg.POSS.ABS(O)	kambi clothes.ABS(O)	julurri-l, wash-NONPAST
	ngayu-ji 1sg.NOM(A)-in.turn	nyungu-nku 3sg-DAT:pt	mayi food.ABS(O)	waju-l. cook-NONPAST
	'She washes my clothes and I in turn cook food for her.'
(400)	Jalhu-naku	naanya	kuni-ny	naayu-ji

(400) Jalbu-ngku nganya kuni-ny, ngayu-ji woman-ERG:pt(A) I sg.ACC(O) hit-PAST Isg.NOM(A)-in.turn

nyunguny kuni-ny. 3sg.ACC(O) hit-PAST 'The woman hit me and I in tum hit her [back].'

4.4 Subordinate clauses

A clause is understood to be subordinate to another clause if it cannot be uttered independently; in other words, a subordinate clause must be directly adjoined to a main or superordinate clause. (Sometimes a subordinate clause by itself may constitute the answer to a question where the question can be regarded as the main clause, e.g. 'Why do you go to the river? - To catch fish' (purposive clause) or 'Why do you take a stick? - Lest the dog attack me' (precautionary clause).) Functionally, we can distinguish seven different types of subordinate clauses in Kuku Yalanji expressing the following concepts: 1. purposive, 2. precautionary, 3. successive, 4. simultaneous action, 5. causal, 6. prior event, 7. hypothetical reason. Clause types 4-7 are all based on verb + -nya constructions (see §3.2.3.6 and §3.8.4.2 (viii» which display nominal characteristics in some clauses.

Within a basic temporal framework of 'former' and 'latter' event with respect to the tense of the main clause these subordinate clauses can be classified as follows:

former	latter
causal	purposive
prior event	precautionary
	successive
	hypothetical reason

'Simultaneous', as the label implies, has the same time reference as the main clause. The preponderance of 'latter event' subordinate clauses is interesting in view of the fact that Kuku Yalanji does not distinguish between present and future in its verbal tense inflections. A variety of subordinate clauses which clearly refer to a future event in the framework of discourse are an efficient means of filling this gap.

Subordinate clauses may or may not share common NPs with the main clause. Deletion possibilities of co-referential NPs in subordinate clauses are very free indeed, as long as the resulting sentence remains unambiguous. In the absence of strict syntactic rules that require or forbid NP deletion, any deleted NP in a subordinate clause is understood to be coreferential with the semantically and contextually most plausible NP in the main clause. If there is an ambiguity resulting from deletion, the preferred reading with regard to the coreferentiality of an omitted NP differs somewhat between the various types of subordinate clauses. Furthermore, not all subordinate clauses have the same range of possible coreferential NPS; for instance, the co-referentiality scope of a purposive construction is much larger than that of a hypothetical reason clause.

Some subordinate clauses may be interpreted as relative clauses, which will be discussed separately in §4.4.5.

4. 4.1 Purposive

A verb with purposive inflection -nkV (see §3.8.4.2 (v» may occur as the only verbal constituent of a main clause as well as in a subordinate clause. In a main clause a purposive verb implies the need or obligation to do something, arising out of a certain necessity, for example:

• (401) Ngayu mayi waju-nku. Dakuy. 1 sg.NOM(A) food.ABS(O) cook-PURP hungry 'I'd better cook some food. [I am] hungry. '
• (402) Ngayku balji kadaba-ny. Ngayu balji 1sg.POSS.ABS(S) bag.ABS(S) break-PAST 1sg.NOM(A) bag.ABS(O)

jirakal balka-nka. new.ABS(O) make-PURP 'My bag is broken. I have to make a new bag. '

A complex sentence with a purposive subordinate clause32 indicates that the action referred to in the main clause is performed so that the action described in the subordinate clause may take place, for instance 'X does Y in order to do Z'. In this function a purposive verb resembles a noun in dative case as 'goal of a purposeful action' (see §4.1 .4.4 (ii)). Purposive verbal inflection and dative nominal inflection are of course also very similar in form.33 Consider:

• (403) a. Yubal wuju mana milkul-ku. 2du.NOM(A) grass.ABS(O) get.IMP soup-DAT 'You two get some grass for the soup!' (i.e. for eating soup; this type of grass was made into a little brush to soak up and eat soup or honey with)
  • b. Yubal wuju mana, milkul nuka-nka. 2du.NOM(A) grass.ABS(O) get.IMP soup.ABS(O) eat-PURP 'You two get some grass for eating the soup!'

A dative noun in this environment indicates that its referent will be subjected to a contextually (and culturally) implicit action, like 'eating' in (403a). A purposive verb form, on the other hand, clearly identifies the type of action that is anticipated as in (403b).

A purposive clause may also imply a normal but unplanned consequence, which arises out of the main clause action, as for instance:

(404)	Karrkay	jurrki-ji-y,	balka-ji-nka.
	child.ABS(S)	move-REF-NONPAST	make-REF-PURP
	'The child moves and as a result gets born.' (i.e. moves down in the womb)

A purposive clause follows the main clause and virtually any co-referential NP may be deleted in the subordinate clause. Obligatory deletion applies to common S or A NPs, i.e. the grammatical subject, and a common 0 NP. Deletion of other co-referential NPs is preferred but will not apply where this renders the sentence ambiguous. Where this is the case, an anaphoric pronoun may be used.

Examples of deletion of NPs in the purposive clause which are co-referential with the S NP in the main clause:

³² See R. Hershberger (1970:818-819).

³³ Verbal and nominal 'purposive' inflections resemble each other in form in many Australian languages; see Capell ( 1 956:77); Dixon (1 980:458).

S = S' ³⁴

• (405) lalbu-jalbu bunda-ny, balka-wa-nka. woman-REDUP.ABS(S) sit-PAST speak-RECIP-PURP 'The women sat down to have a chat.'
• S = A'
  • (406) Warru dunga-ny wawubaja-nga, kuyu dama-nka. yg.man.ABS(S) go-PAST river-LOC fish.ABS(O) spear-PURP 'The young man went to the river to spear fish. '

Further examples are in (41 9) and (420).

S = O'

• (407) lalbu wandi-ny, dunyu-ngku nyaji-nka. woman.ABS(S) rise-PAST husband-ERG:pt(A) see-PURP 'The woman stood up so that [her] husband could see her.'
• S = oblique object' (and oblique object = A')
  • (408) Nyulu jawun-karra-nda kunjaji-ny, mayi daji-nka. 3sg.NOM(S) friend-KPL-LOC:pt ask.for-PAST food.ABS(O) give-PURP 'He asked [his] friends to give [him] some food. '

Here the intransitive subject of the main clause is also understood to be the recipient of the food referred to in the subordinate clause.

Examples of deletion of NPs in the purposive clause which are co-referential with the A NP in the main clause:

A = A' (and O = O')

(409) Ngamu-ngku kuyu waju-l, nuka-nka. mother-ERG:pt(A) fish.ABS(O) cook-NONPAST eat-PURP 'Mother is cooking fish in order to eat [it].'

Deletion of the A NP in the purposive clause implies that it is mother who will eat the fish and not someone else.

A = S'

(41 0) Babi-ngka yangka nuka-l, ngulkurr-ma-nka. FM-ERG:pt(A) green.ant.ABS(O) drink-NONPAST good-INCHO-PURP 'Grandmother drinks green ant Guice) in order to get better.'

Another example of A = S' can be seen in (416).

A = O'

(41 1 ) Ngayu juku duda-ny, bama-ngka nyaji-nka. I sg.NOM(A) tree.ABS(O) beat-PAST Aborigine-ERG:pt(A) hear-PURP 'I beat (on) the tree so that the people can hear [me].'

Here the omitted object in the purposive clause could of course also refer to the noise produced by the hitting of the tree. For this reason it would be preferred to include the 0 NP

³⁴ ' denotes the constituent of the subordinate clause.

nganya 'me' in the purposive clause, although the above version with the given translation is possible.

A = oblique object' (and O = A')

(412) Ngayu jalbu kunja-ny, bulkiji daji-nka. 1sg.NOM(A) woman.ABS(O) summon-PAST clam.ABS(O) give-PURP 'I summoned/ordered the woman to give [me] some clams.'

Examples of deletion of NPs in the purposive clause which are co-referential with the 0 NP in the main clause:

O = O'

(413) Babarr-angka kurmun balka-l, ngamu-ngku eZ-ERG:pt(A) bag.ABS(O) make-NONPAST mother-ERG:pt(A) jarrakulba-nka. carry-PURP 'Elder sister is making a bag for mother to carry.'

Another example of O = O' is in (409).

O = S'

(414) Buliman-angka warru kunja-ny, kada-nka. police-ERG:pt(A) yg.man.ABS(O) summon-PAST come-PURP 'The police summoned the young man to come.'

O = A'

(415) Babi-ngka maral balka-ny, ngawa kuji-nka. FM-ERG:pt(A) girl.ABS(O) speak-PAST baby.ABS(O) look.after-PURP 'Grandmother told the girl to look after the baby.'

Another example of O = A' is in (412).

o = oblique object'

(416) Ngayu kararr ngara-l, bunda-nka. 1sg.NOM(A) sheet.ABS(O) spread-NONPAST sit-PURP 'I spread a sheet to sit on. '

Examples of deletion of purposive clause NPs which are co-referential with the oblique object NP of the main clause:

oblique object = s'

(417) Yaba-ngka jinkurr-undu yalama-ny, warri-nka. eB-ERG:pt(A) yZ-LOC:pt say-PAST run-PURP 'Elder brother told [his] younger sister to run.'

oblique object = A'

(418) Jalbu-ngku manda-nda yalama-ny, woman-ERG:pt(A) niece-LOC:pt say-PAST fire.ABS(O) burn-PAST 'The woman told [her] niece to light a fire.' baya waju-nku.

Another example is in (408).

oblique object = O' (and S = A')

(41 9) Nyulu bayan-ba jirakal-ba dunga-y, nuri-l-nuri-nka. 3sg.NOM(S) house-LOC new-LOC gO-NONPAST peep-l-REDUP-PURP 'He goes to the new house to have a peep [at it].'

oblique object = oblique object' (and S = A')

(420) N gayu maja-nda dunga-y, baba-ji-nka. I sg.NOM(S) bOSS-LOc:pt go-NONPAST ask-PURP 'I go to the boss to ask [him].'

It will have been noticed that all oblique objects in the examples are in locative case. Presumably there is no syntactic reason why the same deletion possibilities should not apply to oblique objects in other cases; however, it is very difficult to establish a semantically plausible environment in which other than locative oblique objects are co-referential with another NP (and 1 have not succeeded in doing so).

Of all the possibilities of co-referential NPs illustrated above some of course are more frequent than others, corresponding to the relative frequency of NPs in various functions. Co-referentiality of the S or A NP and of the 0 NP is the most frequent, while it is relatively uncommon for an oblique object to be co-referential with an NP in the same or a different function. The latter is in fact limited to verbs of utterance, for instance indirect speech as in (408), (4 1 7) and (4 1 8), the verb daji-l 'give', as in (408) and (4 1 2), and local expressions as in (41 6), (41 9) and (420).

It is not uncommon for both the A and 0 NP to be deleted in a purposive clause as for instance in (409). With such a large range of deletion possibilities certain ambiguities may arise. Consider the following sentences:

• (42 1) Kami-ngka karrkay wundi-l, wawurrwawurr-ma-nka. FFIMM-ERG:pt(A) child.ABS(O) take-NONPAST happy-INCHO-PURP 'Granny takes the child so that [he/she/it] be happy.'
• (422) Kami-ngka karrkay wundi-l, baya FF/MM-ERG:pt(A) child.ABS(O) take-NONPAST firewood.ABS(O) gather-PURP 'Granny takes the child to gather firewood.' muja-nka.

These sentences were considered somewhat ambiguous by informants in that it appeared unclear who would be happy and who would collect the firewood. However, preference was given to the interpretation in which it was "granny" in both cases, which is in keeping with the obligatory deletion of the co-referential S/A NP in the purposive clause. To make it clear that the child is meant to be the S or A NP of the purposive clause it is sufficient to include an anaphoric pronoun. For instance:

(423)	Kami-ngka	maral	wundi-ny,	nyulu	kambi
	FF/MM-ERG:pt(A)	girl.ABS(O)	take-PAST	3sg.NOM(A)	clothes.ABS(O)
	nyungu	mani-nka.
	3sg.POSS.ABS(O)	get-PURP

'Granny took the girl so that she (the girl) will get her clothes. '

Nyulu in the purposive clause can only be co-referential with maral, the O NP of the main clause, because co-referential A = A' would require obligatory deletion of A'.

A purposive clause need not share a common NP with the main clause as for example:

(424) Ngayu baya waju-l, 1sg.NOM(A) fire.ABS(O) burn-NONPAST 2sg.NOM(S) warm-INCHO-PURP 'I light a fire so that you get warm.' yundu wumbul-ma-nka.

(425) Dunyu-ngku kamukamu kari nuka-l, manyarr husband-ERG:pt(A) a1cohol.ABS(O) NEG drink-NONPAST wife.ABS(S) wawurrwawurr-ma-nka. happy-INCHO-PURP

'The husband does not drink alcohol so that [his] wife is happy.'

Like nouns in dative case, the purposive fonn of the verb is also required in connection with an adjectival particle (see §3.9.1):

• (426) a. Nyulu wawu minya-ka. 3sg.NOM(S) want meat-DAT 'He wants some meat.'
  • b. Nyulu wawu nuka-nka. 3sg.NOM(S) want eat-PURP 'He wants to eat.'
  • c. Nyulu wawu minya nuka-nka. 3sg.NOM(S) want meat.ABS(O) eat-PURP He wants to eat meat.
• (427) N gayu wawu jana 1sg.NOM(S) want 3pl.NOM(A) turtle.ABS(O) get-PURP 'I want them to get some turtle. ' ngawuya mani-nka.
• (428) a. Ngayu yinil wujurr-ku. ysg.NOM(S) afraid darkness-DAT 'I am afraid of the dark.'
  • b. Ngayu yinil wujurr-bu 1sg.NOM(S) afraid darkness-LOC go-PURP 'I am afraid to go out in the dark.' dunga-nka.

Here the purposive clause is subordinate to a nominal sentence with which it may or may not share a common NP.

Consider also:

• (429) a. Yanya bana nuka-nka. this water drink-PURP 'This is drinking water/water for drinking.'
  • b. Yinya junjuyjunju:y waju-nku. that something/rubbish burn-PURP 'That is rubbish for burning.'

In these examples the purposive verb can probably be understood as an abbreviated subordinate clause in which a non-specific agent is implicit, i.e. 'water for someone to drink', 'rubbish for someone to burn'. The purposive verb is not part of the NP since case marking cannot be applied to both the noun and the verb form; the following sentence with locative inflection is ungrammatical.

(430) *Bulbur bana-nga nuka-nka-rnun-bu dust water-LOC drink-PURP-mun-LOC 'There is dust in the drinking water.'

To express this idea the sentence would have to be paraphrased as something like:

(43 1) Yanya bana nuka-nka, yarnba bulbur bana-nga. this water drink-PURP but dust water-LOC 'This is drinking water, but there is dust in the water. '

4. 4.2 Precautionary

A precautionary verb form, stem + -nyji (see §3.8.4.2 (vi)), is used only in subordinate clauses. It conveys that the action referred to in the main clause is or should be performed in order to prevent the action or event described in the subordinate clause. This can be formalised as 'x does/does not do y lest z happen'.

A precautionary clause follows the main clause. Co-referential NP deletion possibilities are as extensive as those for purposive constructions; however, there is apparently no semantic environment in which an oblique object in the precautionary clause can be coreferential with an NP in the main clause. Again, obligatory deletion in the subordinate clause applies to co-referential S/A NPs, unless the main clause is an imperative as in (433), and to a common 0 NP. Other deletions are optional but preferred. Examples:

S = S'

(432) Nyulu kari janji-l wawubaja-nga, bujil-janji-nyji. 3sg.NOM(S) NEG bathe-NONPAST river-LOC nose-bathe-PRECAUT 'He does not bathe in the river lest [he] drown.'

S = A'

(433) Kari kalka-ji nguju-wuri-y, yundu ngawa kuni-nyji. NEG spear-COMIT funny-dance-IMP 2sg.NOM(A) baby.ABS(O) hit-PRECAUT 'Don't play/fool around with the spear lest you hit the baby!'

If a precautionary construction contains an imperative, which is often the case, the AlS NP is usually stated in the subordinate and not in the main clause, but it may be omitted altogether as in (436) and (44 1 ).

S = O'

(434)	Bama Aborigine.ABS(S)	kari	jikan-ba grass-LOC	kalbay-mba long-LOC	dunga-n-dunga-y, go-n-REDUP-NONPAST
		jarba-bu snake-ERG(A)	bayka-nyji. bite-PRECAUT	'Aborigines don't walk through long grass lest a snake bite [them].'

$$\mathbf{A} = \mathbf{S}'$$

(435) Warru-warru-ngku karnukarnu kari nuka-l-nuka-l, yg.man-REDUP-ERG:pt(A) a1cohol.ABS(O) NEG drink-l-REDUP-NONPAST kuni-wa-nyji. hit -RECIP-PRECA UT 'Young men don't drink alcohol lest [they] fight. '

A = A'

(436) Mayi kari waju, buyun-dama-nyji. food.ABS(O) NEG cook.IMP bad-CAUS-PRECAUT 'Don't cook the food lest you spoil [it] !'

A=o' (ando=A')

(437) Kari kaya kuli-ji karrba, (yununy) bayka-nyji. NEG dog.ABS(O) wild-COMIT.ABS(O) touch.lMP 2sg.ACC(0) bite-PRECAUT 'Don't touch the vicious dog lest [it] bite [you]. '

In the last example it is perfectly clear who will do the biting and who will suffer it. It would therefore be safe to omit both the co-referential A and 0 NP in the subordinate clause. But consider:

(43 8) Babarr-angka karrkay-karrkay kanga-ny, nyunguny eZ-ERG:pt(A) child-REDUP.ABS(O) chase-PAST 3sg.ACC(0) kuni-nyji. Karrkay kuli-baka. hit-PRECAUT child(ren) wild-excess 'Elder sister chases the children away, lest they hit her. The children are a wild lot.'

Here, deletion of both co-referential NPs would result in some doubt as to who hit whom.

O = S'

(439) Ngayu ngawa jarrakulba-l, banbadi-nyji. I sg.NOM(A) baby.ABS(O) carry-NONPAST cry.REDUP-PRECAUT 'I carry the baby lest [it] cry.'

O = A'

(440) Bula jalbu-ngku dubu waJu-ny, kangkal 3du.NOM(A) woman-ERG:pt(A) ghOSt.ABS(O) burn-PAST own.childABS(O) ngaki-nyji-baja. steal-PRECA UT -again

'The two women burnt the ghost lest [he] steal [their] children again.'

Another example of $O = A'$ is in (437).

$O = O'$

(441) Kari diburr kulba, dara-y-mani-nyji. NEG egg.ABS(O) lift.IMP fall-y-CAUS-PRECAUT 'Don't pick up the egg lest you drop [it] !'

A subordinate NP that is co-referential with an oblique object in the main clause may be deleted, but retention in some form is preferred, for example:

(442)	Nyulu	wawu	kari	kaya-nda	dunga-nka,
	3sg.NOM(S)	want	NEG	dog-LOC:pt	go-PURP

(kaya-ngka) / (nyulu) bayka-nyji. dog-ERG:pt(A) / 3sg.NOM(A) bite-PRECAUT 'He doesn't want to go to the dog, lest (the dog)/(it) bite [him].'

Note that in (442) the precautionary clause is subordinated to a purposive clause.

A precautionary construction may also occur without a common NP, but this is not very frequent. For instance:

(443) Baya waju-l-waju, kiway kada-nyji. fire.ABS(O) burn-REDUP-IMP cold.ABS(S) come-PRECAUT 'Keep a fire burning lest it get cold!'

Occasionally, a precautionary clause may be uttered on its own. This is in the nature of a general warning where the addressee is supposed to know what to do to remedy the situation. If someone said:

(444) Yinya kuyu kabu-ma-nyji. that.ABS(S) fish.ABS(O) rotten-INCHO-PRECAUT 'That fish might get rotten. '

the addressee would infer that something ought to be done in order to preserve the fish.

As was mentioned in the previous section, the function of a purposive clause resembles that of a noun in dative case, i.e. 'goal of a purposeful action'. Similarly, in some contexts, the function of a precautionary clause may be compared to that of a dative noun expressing 'general threat for fear of which something is done or not done' (see §4.1 .4.4 (ii), examples (262) and (263)). Again, while the dative noun merely implies a threatening action by its referent, the precautionary clause spells out exactly what the danger is.

4.4.3 Successive

The verb form root + -nyjiku (see §3.8.4.2 (vii)) in a subordinate clause indicates that this event follows closely after the event referred to in the main clause: 'X does Y shortly before Z happens'.35 The two events are seen in a temporal, and to some extent logical, but not a causal, relationship. It appears that the event in the subordinate clause can be regarded as a temporal reference point with respect to which the timing of the main clause is established. This interpretation is supported by the fact that the subordinate clause action may be unrealised, as in:

(445) Ngayu jinbal-da-ku jarba kuni-ny, nganya I sg.NOM(A) quick-INST-ADV snake.ABS(O) kill-PAST I sg.ACCCO) bayka-nyjiku. bite-sUCC 'I quickly killed the snake before [it] might bite me. '

Such a sentence would serve as an answer to the question 'when did you kill the snake?' (Snakes are normally expected to bite and this question would more likely to prompted by concern for the snake killer than by the desire to know whether it happened before or after

³⁵ R. Hershberger (1 979:76-77) uses the term "Prior Time Clause" for this type of subordinate clause. This is a misnomer since the subordinate clause actually refers to the following event and the main clause to the preceding one.

1 70 Chapter 4

breakfast.) On the other hand, a precautionary clause would be used, in my informant's words, "when a fellow ask you: 'What did you want to kill that snake for?'"

Co-referentiality in successive constructions appears to be restricted to core syntactic NPs. Of these, common NPs with the same grammatical function, i.e. $A/S = A'/S'$ and $O = O'$ , are obligatorily deleted in the subordinate clause. Examples:

$S = S'$

(446) N gayu wunana-ny, jilba-dunga-nyjiku. I sg.NOM(S) lie.REDUP-PAST walk-go-succ 'I had a rest before going for a walk.'

$S = A'$

(447) lalbu mara julurri-ji-y, mayi woman.ABS(S) hand.ABS(S) wash-REF-NONPAST food.ABS(O) damba dingka-nyjiku. damper.ABS(O) knead-SUCC 'The woman washes [her] hands before kneading damper.'

$A = S'$

(448) Maral-angka kambi bangka-l, dunyu-ndu girl-ERG:pt(A) clothes.ABS(O) stock.up-NONPAST husband-LOC:pt lie-SUCC 'The girl stocks up clothes before getting married.' wuna-nyjiku.

$A = A'$

(449) Nyulu bubu kida-l-kida-l, bayan 3sg.NOM(A) ground.ABS(O) scrape-l-REDUP-NONPAST camp.ABS(O) ngara-nyjiku. set.up-SUcc 'She sweeps the ground before setting up camp.'

$O = O'$ (and $A = A'$ )

(450) lalbu-jalbu-ngku marra dinda-l, duda-nyjiku. woman-REDUP-ERG:pt(A) zamia.nut.ABS(O) roast-NONPAST pound-SUCC 'The women roast the zamia nuts before pounding them.'

Note that successive subordinate constructions with common 0 NP but different A NP are not favoured, although possible. Sentence (450) with different agents would rather be rendered as:

(45 1) lalbujalbungku marra dindal, maral-maral-angka marra girl-REDUP-ERG:pt(A) zamia.nut.ABS(O)

duda-l-da.

pound-NONPASf-TEMPl

'The women roast the zamia nuts, then the girls pound the nuts.'

where marra can be optionally deleted in the second clause.

Successive constructions may of course have co-referential NPs in different functions, but deletion is not liked in these cases. Pronoun reduction is much preferred, even if the sentence is quite unambiguous, as for instance in (445) where $A = O'$ and $O = A'$ . Examples of $O = A'$ or $S'$ are:

• (452) Ngayu kaya karrba-ny, nyulu ngayku I sg.NOM(A) dog.ABS(O) grab-PAsT 3sg.NOM(A) I sg.POSS.ABS(O) minya bangkarr nuka-nyjiku. meat.ABS(O) steak.ABS(O) eat-SUCC 'I grabbed the dog before it ate my steak.'
• (453) Jana kadar nyaji-ny, nyulu warri-nyjiku. 3pLNOM(A) wallaby.ABS(O) see-PAST 3sg.NOM(S) run-SUCC 'They saw the wallaby before it ran away. '

However, successive constructions may also have no common NP as the following examples show:

• (454) Dikal banbadi-y, kaba dara-nyjiku, kababinal. bird.ABS(S) cry.REDUP-NONPAST rain.ABS(S) fall-SUCC rainbird 'A bird cries before rain falls, [that is] the rainbird.'
• (455) Ngayu dunga-ny, muyar kada-nyjiku. I sg.NOM(S) go-PAST wind.ABS(S) come-SUCC 'I went before the wind came.'
• (456) Nganjin bibikarrbi wunana-ny, ngayku 3pLexc.NOM(S) Helenvale lie.REDUP-PAST I sg.POSS.ABS(O)

nganjan wundi-nyjiku waybala-ngka Palm Island. father.ABS(O) take-succ white.man-ERG:pt(A) Palm.lsland 'We were living at Helenvale before white people took my father to Palm Island.'

Successive constructions are not all that often used in Kuku Yalanji. Those without coreferential NPs are somewhat more common than the others. Where the same actor (and object) is involved in a number of successive actions, co-ordinated topic chains are by far more frequent than subordinate constructions.

4.4. 4 Verb + -nyV constructions

Subordinate clauses containing verb + -nyV forms have various functions. They may indicate 'simultaneous action', 'causal', 'prior event', and 'hypothetical reason'. In all but causal clauses the verb + -nyV form has nominal characteristics in that it allows regular neutral case inflections. (-nyV will be glossed as SUB in all functions.)

4.4.4.1 Simultaneous action

If two, or more, actions are performed simultaneously by the same or different actors, a subordinate clause with verb + -nyV may be used.³⁶ (A more loose expression of simultaneous action may be effected by sentence co-ordination; see §4.3. 1 .) The verb form

³⁶See R. Hershberger's (1 979:79-80) discussion of "Participle Clause". This term was presumably chosen by R. Hershberger because of the case-marking possibilities of verb + -nyV forms. Note that her example of a "Participle Clause" as "Subject of a causative verb" has not been confirmed by my informants. Causal subordinate clauses with verb + -nyV are discussed in the following subsection.

1 72 Chapter 4

in a simultaneous action subordinate clause is often reduplicated because an action would normally have to be ongoing to allow another action to concur. But there are occasions where a non-reduplicated form may be used to indicate simultaneity at a particular moment, as for instance 'I saw the man at the moment that he killed the snake'.

Where two different actors are involved, the main and subordinate clause may have the following common NPs: O = S' or A', oblique object = S' or A'. A sentence with O = S' or A' has the underlying structure:

[A NP 0 NPI tr Verbh : [A/S-NPI (0 NP) Verb] L'

The co-referential NP, marked by subscript b may be deleted in either clause in surface structure and the clause Σ' verb has the surface form: (reduplicated) stem + -nyV. Main and subordinate clause are separated by slight intonation rise on the main-clause verb followed usually by a short pause. This is important for the placement of a co-referential O = S' NP, which shows no formal distinction in form if it is a noun. (In the written presentation the clause boundary is indicated by a comma as in the previously discussed subordinate constructions. )

According to where the co-referential NP is deleted, the surface form has a somewhat different interpretation. Consider the following sentences where the common NP is deleted in the subordinate clause:

• (457) Ngayu yununy nyaji-ny, mayi yurmbi-I-yurmbi-nya l sg.NOM(A) 2sg.ACC(0) hear-PAST food.ABS(O) chew-l-REDUP-SUB jangku-ji. chewing.noise-COMIT 'I heard you chewing food noisily. '
• (458) Bama-ngka Janzman nyaji-I-nyaji-ny, gold Aborigine-ERG:pt(A) Chinese.ABS(O) see-l-REDUP-PAST gold.ABS(O)

baka-I-baka-nya. dig-l-REDUP-SUB 'The Aborigine watched the Chinese digging for gold. '

• (459) Wara-ngka warru nyarrma-ny, ngawuya groper-ERG:pt(A) yg.man.ABS(O) grab.in.passing-PAST turtle.ABS(O) dama-l-dama-nya. spear-REDUP-SUB 'The groper grabbed in passing the chap spearing turtles. '
• (460) Buliman-angka nyunguny kuni-ny, wanarri-nya. police-ERG:pt(A) 3sg.ACC(0) hit-PAST run.REDUP-SUB 'The policeman shot him while he (i.e. the victim) was running.'

The following is an example where 0 is co-referential with the S of a subordinate antipassive:

(46 1) Ngayu kadar kungkunbunga-l-bunga-l, jikan-ba l sg.NOM(A) wallaby.ABS(O) watch-l-REDUP-NONPAST grass-LOC nuka-nuka-ji-nya. eat-REDUP-ITR-SUB 'I am watching the wallaby feeding on grass.'

In the above sentences the deletion pattern seems to single out the 0 NP. This could be appropriately expressed by a relative clause in English, e.g. (458) 'he watched the Chinese who was digging for gold' or (459) 'the groper grabbed the chap who was spearing turtles'. The preferred word order in the main clause of such sentences is AOV, as shown above, although AVO has also been heard occasionally.

Now compare the following sentences where the 0 NP is deleted in the main clause and the co-referential A or S NP retained in the subordinate clause. Here the emphasis lies on the action described in the subordinate clause as illustrated by the translations:

(462) Bama-ngka nyaji-l-nyaji-ny, janiman-angka gold Aborigine-ERG:pt(A) see-l-REDUP-PAST Chinese-ERG:pt(A) gold.ABS(O) baka-l-baka-nya. dig-l-REDUP-SUB 'The Aborigine watched while/how the Chinese was digging for gold.' (463) Ngayu nyaji-ny, nyulu bulki yaka-l-yaka-nya. I sg.NOM(A) see-PAST 3sg.NOM(A) bullock.ABS(O) cut.up-l-REDUP-SUB 'I saw how he cut up the bullock [carcass].'

(464) Maral-angka nyaji-l-nyaji-ny, bama girl-ERG:pt(A) see-l-REDUP-PAST Aborigine.ABS(S) corroboree wuri-n-wuri-nya. dance-REDUP-SUB warrma 'The girl watched while/how the Aborigines danced a corroboree.'

(465) Karrkay-angka nyaji-ny, jarba kuni-ji-nya. child-ERG:pt(A) see-PAST snake.ABS(S) kill-ITR-SUB 'The child saw the snake being killedlhow the snake was killed. '

In the last example the deleted main clause 0 is co-referential with the S of an agentless passive construction.

Just as a retained 0 NP is inseparable from the main clause, as illustrated previously, a retained subordinate S/A NP must be contiguous to the rest of the subordinate clause.

Note that deletion of the main clause 0 NP is apparently only possible if the main verb is a verb of perception such as nyaji-l 'see, hear' or kungkun-bunga-l 'watch closely'. This suggests that the entire subordinate clause is seen in object function, whereas a subordinate clause in which the AlS NP is deleted seems to relativise the 0 NP of the main clause. (This will be discussed in more detail in §4.4.5.) If the subordinate clause in its entirety is in 0 function one may argue that the clause as a whole takes absolutive case clause finally on the verb + -nyV form. This cannot be verified since absolutive has zero realisation; however, evidence in (466b) and (467b), scant as it is, may support this theory. Simultaneous action constructions in which the main clause oblique object is co-referential with the subordinate S/A NP are quite rare, but consider the following examples:

(466) a. N gayu wulman-anda dunga-y, yinyay bundanda-nya. 1 sg.NOM(S) old.man -LOC:pt go-NONP AST there sit.REDUP-SUB 'I go to the old man (who is?) sitting there.'

b. N gayu dungay wulmananda bundanda-nya-mun-bu. sit.REDUP-SUB-mun-LOC

'I go to the sitting old man.'

(467) a. lana warri-y(,) jalbu-ndu(,) mayi 3pl.NOM(S) run-NONPAST woman-LOC:pt food.ABS(O) cook-SUB 'They run to the woman (who is?) cooking food.' waju-nyu.

b. lana warriy jalbundu mayi waju-nyu-mun-bu. cook -SUB-mun-LOC

'They run to the food-cooking woman.'

It appears that in the a. sentences the subordinate S/A NP is deleted, and the remaining subordinate clause modifies the oblique object. But note that there is no case agreement with the oblique object. In the b. sentences it seems that the entire subordinate clause fulfils oblique object function since the whole clause takes locative marking. Verb + -nyV in this context can in fact be interpreted as a kind of adjective. (Note that the -nyV stem takes neutral case marking following the stem-forming suffix -mun-, whereas 'normal' adjectives take potent inflection in connection with a potent-inflected noun.) However, it remains unclear what (if anything) is deleted where in these examples. The few sentences available do not shed any light on this.

Where simultaneous actions are performed by the same actor, S/A is co-referential with S'/A'. Unlike sentences with common O = S/A NP', which can delete the common NP in either the main or subordinate clause, same-actor sentences apparently allow deletion only in the subordinate clause. This assumption is based on evidence from sentences with crossfunctional common NPs, i.e. S = A' or A = S'. Where S = S' or A = A', it is not discernable which NP is deleted, not even by intonation. The truncated subordinate clause may be embedded in or follow the main clause.

The form of the subordinate verb in these sentences depends largely on whether the main clause is intransitive or transitive. If the main clause is transitive, the subordinate verb optionally takes neutral ergative inflection if the subordinate clause is embedded, and it must do so if the subordinate clause follows the main clause. Thus from an underlying structure:

[A NPI 0 NP verbIJ L [s or A NPI (0 NP) verb2] L'

we obtain either

[A NP O NP [(O NP) verb2-nyV (-ERG)] Σ' verb1] Σ [(0 NP) verb2-nyV-ERG] L'

For example:

A = S'

or

(468)	Jalbu-ngku woman-ERG:pt(A)	bundanda-nya-mun-du sit.REDUP-SUB-mun-ERG	ngayku 1sg.POSS.ABS(O)
	kuku language.ABS(O)	daji-l-daji-ny. give-l-REDUP-PAST 'The woman used to give (teach) me language while sitting down.' (the woman used to sit while teaching, she is not sitting now)
(469)	Jana 3pl.NOM(A)	warru-ngku yg.man-ERG:pt(A)	kinka-n-kinka-nya kamukamu play-n-REDUP-SUB alcohol.ABS(O)
	nuka-l-nuka-l. drink-l-REDUP-NONPAST

'The young men drink alcohol while playing (cards).'

A=A'

(470) lalbu-jalbu-ngku marra duda-I-duda-nya-mun-du woman-REDUP-ERG:pt(A) zamia.nut.ABS(O) pound-I-REDUP-SUB-mun-ERG dirka-bunga-l.

fine-CAUS-NONP AST

'The women make the zamia nut (into a) fine (powder) pounding it.'

(47 1) Dingkar-angka kalka dudu kulji-bu kida-I-kida-nya man-ERG:pt(A) spear.ABS(O) blunt-ABS stone-INST scrape-l-REDUP-SUB

nyurrba-l.

sharpen-NONPAST

'The man sharpens the blunt spear scraping it with a stone.'

In the above sentences the subordinate clauses are embedded in the main clause and ergative inflection on the verb + -nya form is optional. Now consider:

A = S'

(47

(472) Maral-angka girl-ERG:pt(A) Aborigine.ABS(O) see-l-REDUP-PAST bama nyaji-I-nyaji-ny,

wuri-n-wuri-nya-mun-du.

dance-n-REDUP-SUB-mun-ERG

'The girl watched the Aborigines while [she was] dancing. '

A = A'

(473)

N gamu-ngku ngawa mother-ERG:pt(A) baby.ABS(O) look.after-l-REDUP-NONPAST food.ABS(O) kuji-I-kuji-l, mayi

nuka-I-nuka-nya-mun-du. eat-l-REDUP-SUB-mun-ERG 'Mother looks after the baby while [she is] eating. '

Here the subordinate clauses follow the main clause and ergative inflection is obligatory.

Because of the ergative inflection on the verb + -nyV forms, the subordinate clause may be regarded as an attribute to the main clause NP. In some cases the subordinate clause could also be interpreted as referring to the manner in which something is done, as for instance in (470), and (47 1 ), which would justify the term "participle clause" used by R. Hershberger (1 979:79). Where A NP and attributive subordinate clause are contiguous, the need for ergative inflection is not so great, since the sentence cannot be interpreted in any other way than A = A'/S'. If, however, the subordinate clause stands sentence-finally, it must be clearly identified as 'belonging to' the main A NP, because in this position it could just as well be coreferential with the main 0 NP. For instance, compare (472) with (464); without the ergative inflection in (472) it would be ambiguous who is doing what in either sentence. Or take (460), which implies that the victim was running. To indicate that the policeman was running, the subordinate verb must take ergative inflection:

(474) Bulimanangka nyunguny kuniny, wanarri-nya-mun-du policeman-ERG:pt(A) 3sg.ACC(O) hit.PAST run-REDUP-SUB-mun-ERG 'The policeman shot him while he (the policeman) was running.'

If the main clause is intransitive and $S = S'/A'$ , the verb + -nyV form is not further inflected, or one may say that the subordinate verb+ -nyV form is in absolutive case for which inflection is zero. Examples of embedded subordinate clauses where $S = S'$ and $S = A'$ are:

$S = S'$

• (475) Yinya yarraman janjana-nya warngku-wunana-y. that.ABS(S) horse.ABS(S) stand.REDUP-SUB.ABS(S) sleep-lie.REDUP-NONPAST 'That horse sleeps standing up. '
• (476) Nyulu juku-ngu daka-l-daka-nya dara-ny. 3sg.NOM(S) tree-LOC climb-l-REDUP-SUB.ABS(O) fall-PAST 'Climbing on the tree, he fell.'

$S = A'$

• (477) Nyulu jarba nyanma-l-nyanma-nya yirrka-ny. 3sg.NOM(S) snake.ABS(O) hold.on-l-REDUP-SUB.ABS(O) call.out-PAST 'Holding on to the snake, he called out.'
• (478) Bula warru nambarr nanda-l-nanda-nya 3du.NOM(S) yg.man.ABS(S) bamboo.ABS(O) plant-l-REDUP-SUBABS(S) dunga-n-dunga-ny. go-n-RED UP-PAST 'The two chaps went along planting bamboo (all the way).'

Examples of adjoined subordinate clause where $S = S'$ and $S = A'$ are:

'The woman was chatting ignoring [her] husband.'

$S = S'$

(479)	Karrkay child.ABS(S)	wanarri-y, run.REDUP-NONPAST	banbadi-nya. cry.REDUP-SUB.ABS(S)
	'The child runs crying.'
(480)	Dingkar man.ABS(S)	bambay sick.ABS(S)	warngku-wuna-y, burrki-n-burrki-nya. sleep-lie-NONPAST fidget-n-REDUP-SUB.ABS(S)
	'The sick man sleeps tossing and turning.'
S = A'
(481)	Bula 3du.NOM(S)	maral baral-ba girl.ABS(S) road-LOC	kankada-y, come.REDUP-NONPAST
	mayngku mango.ABS(O)	nuka-l-nuka-nya. eat-l-REDUP-SUB.ABS(S)
	'The two girls are coming on the road eating mangoes.'
(482)	Nyulu 3sg.NOM(S)	jalbu balkawa-ny, dunyu woman.ABS(S) chat-PAST husband.ABS(O)	bayja-nya. ignore-SUB.ABS(S)

4.4. 4.2 Causal

In causal subordinate clauses37 the verb + -nyV form does not allow case inflections and can therefore not be regarded as a nominal as in the other subordinate clauses containing this

³⁷ See R. Hershberger (1970:8 1 9-820) 'Causative Sentence'.

form. (Note that a noun in causal function takes ablative; see §4. 1 .4.4 (iv).) The causal action is understood to be completed and prior in time to the main verb. Of course, a simultaneous action may also sometimes imply a logical cause, for example:

(483) Ngayu muku kaka, baya muja-l-muja-nya. I sg.NOM(S) back.ABS pain.ABs firewood.ABS(O) gather-l-REDUP-SUB.ABS(S) 'I have a sore back while/from gathering firewood.'

Here, the gathering of firewood is presumably also the cause of the backache, but emphasis in this case lies in the coinciding of both events as compared to:

(484) Ngayu muku kaka, baya muja-nya yilayku. I sg.NOM(S) back.ABS pain.ABS firewood.ABS(O) gather-sUB yesterday 'I have a sore back from gathering firewood yesterday. '

Here the past firewood collection is the cause of the backache which still persists at the time of utterance.

A causal clause may precede or follow the main clause. Treatment of common NPs in main and causal clause is somewhat different to that in previously discussed subordinate constructions. The following common NPs are possible:

$S$ or $A = S'$ or $A'$ O = O' or S' or A'

But consistent deletion of the subordinate NP occurs only where S = S' or A', for instance:

S = S'

(485) Janji-nya, nyulu bajaburray. bathe-SUB 3sg.NOM(S) tired 'He is tired from bathing. '

S = A'

(486) Mayi durrba-nya, nyulu manu jayba-ny. food.ABS(O) gobble-sUB 3sg.NOM(S) throat.ABS(S) get.stuck-PAST 'He choked from gobbling down food.'

Another example of S = A' is in (483).

Where a causal clause NP is co-referential with the A or 0 NP of the main clause, the use of a pronoun is preferred for clarity, although deletion is possible. For example:

A = S'

(487) Buliman-angka nganya kari nyaji-ny, (nyulu) warngku-wuna-nya. police-ERG:pt(A) 1sg.ACC(O) NEG see-PAST 3sg.NOM(S) sleep-lie-SUB 'The policeman did not see me because [he] was asleep.'

Without the anaphoric pronoun this sentence would be ambiguous as to who was asleep.

$$\mathbf{A} = \mathbf{A}"\text{ and }\mathbf{O} = \mathbf{O}"$$

(488)	Babarr-angka eZ-ERG:pt(A)	karrkay child.ABS(O)	nubi-l-nubi-l, search.for-l-REDUP-NONPAST	(nyulu) 3sg.NOM(A)
	(nyunguny) 3sg.ACC(O)	nubiymani-nya. lose-SUB	'Elder sister is searching for the child because [she] lost [it].'

O = S'

(489) Bula karrkay wundi-ny, (nyulu) milka-bujarr-ma-nya. 3du.NOM(S) child.ABS(O) take-PAST 3sg.NOM(S) ear-cold/raw-INCHO-SUB 'The two took the child (with them) because (it) felt lonely. '

O = A'

(490) N ganjan-angka kangkal murrukanga-l, father-ERG:pt(A) own.child.ABS(O) scold-NONPAST 3sg.NOM(A) kaya jabi-nya. dog.ABS(O) kick-SUB (nyulu) 'Father is scolding [his] child because [it] kicked the dog.'

On first sight (490) is still ambiguous since the anaphoric pronoun could just as well refer to 'father', and the same would apply if the agent in (489) were also third person singular. In such instances the context of the utterance will disambiguate the sentence, or the most plausible interpretation would apply. For instance it would be unlikely that father is scolding his child because he (the father) kicked the dog.

4.4.4.3 Prior event

The previous section discussed expression of events that are prior in time as well as causal with relation to another event. In this section we will look at subordinate clauses which indicate events that precede another event but do not have any causal effect. These can be generalised as 'X does Y after having done Z'. The prior action is expressed by wawu + verb + -nyV + Ablative. (For temporal proclitic wawu- see §3.10.1). As illustrated in §4.1.4.4 (iv), ablative indicates, among other things, a (locational or) temporal starting point which has been left behind by the actor, which means that the preceding action is thought to be completed.

A prior-event clause with verb + -nyV form may precede or follow the main clause, but it is only possible if the following NPs are co-referential:

$$\begin{array}{l} \mathsf{S} = \mathsf{S}'\ \bullet = \mathsf{S}' \text{ or } \mathsf{A}' \text{ (if also: } \mathsf{A} = \mathsf{O}') \end{array}$$

The co-referential NP is deleted in the subordinate clause.

S = S'

(49 1) Wawu-janji-nya-muny, jana wunana-y. TEMP4-bathe-SUB-ABL 3pl.NOM(S) lie.REDUP-NONPAST 'After bathing, they have a rest.'

O = S'

(492) Nganjin jalbu nyaji-ny, wawu-wala-nya-muny. Ipl.exc.NOM(A) woman.ABS(O) see-PAST TEMP4-enter-SUB-ABL 'We saw the woman after [she] had come in.'

O = A' and A = O'

(493) Dingkar-angka jinkalmu kuni-ny, wawu-bayka-nya-muny. man-ERG:pt(A) brown.snake.ABS(O) kill-PAST lliMP4-bite-sUB-ABL 'The man killed the Brown Snake after [it] had bitten [him].'

Where main and subordinate clause share a common NP in S and A function respectively, the subordinate A NP is deleted, but it is the subordinate 0 or oblique object NP which takes ablative marking while the verb remains unchanged, as for example:

• (494) Ngayu wunana-ny, wawu-kambi-muny julurri-ny. I sg.NOM(S) lie.REDUP-PAST lliMP4-clotheS-ABL wash-PAST 'I had a rest after washing clothes. '
• (495) Nyulu jilba-dunga-ny, wawu-jukijuki-muny mayi daji-ny. 3sg.NOM(S) walk-go-PAST lliMP4-chicken-ABL food.ABS(O) give-PAsT 'She went for a walk after having fed the chickens.'

Note that the oblique object in the subordinate clause in (495) does not receive locative case marking before the ablative inflection as would be required in a simple sentence 'she gave food to the chickens (-LOC)' (see §4.1 .4.4 (iii) example (272)).

It could be argued that wawu- -muny in the subordinate clause attaches to the first word and that the subordinate verb is nominalised (by -nyV) only if it must bear case inflection. As a rule, ablative marking is attracted by nouns away from a verb + -nyV form. Therefore, if an event can possibly be expressed by a nominal, this must be used to indicate the prior event; a verb + -nyV form is unacceptable if an alternative noun is available. For instance:

(496)	Warru-ngku yg.man-ERG:pt(A)	kaya dog.ABS(O)	karrba-ny grab-PAST	wawu-ngangkurr-muny / TEMP4-barking.noise-ABL
*wawu-ngangkurri-nya-muny TEMP4-bark-SUB-ABL 'The young man grabbed the dog after it had barked (lit. after the barking)'.
(497)	Nyulu 3sg.NOM(A)	bana water.ABS(O)	nuka-ny drink-PAST	wawu-jilba-muny / TEMP4-walk-ABL

*wawu-dunga-nya-muny. lliMP4-go-SUB-ABL 'She drank some water after the walk.'

Apparently a prior-event subordinate construction is not possible if a main clause A NP is co-referential with a subordinate A or S NP, presumably because co-referential O = S'/A' and A = S'/A' could be confused (see (492)). To express that the referent of an A NP has performed another action previously, this action may be indicated by a noun in ablative case, as for instance in (497), or two clauses are simply conjoined, using the general expression wawu-yinya-muny 'after that'.

4. 4. 4.4 Hypothetical reason

Purposive and precautionary (see §4.4.1 and §4.4.2) constructions express real reasons for which something is done. These reasons are real because the events indicated by purposive or precautionary are definitely expected to happen. Kuku Yalanji has also a means of

180 Chapter 4

expressing a hypothetical reason because of which something else is done. This may be generalised as 'X does Y just in case Z happens'. The hypothetical reason 'Z' is expressed by verb + -nyV + -Dative (cf. function of Dative as indicator of general reason §4.1.4.4 (ii)). For example:

(498)	Jana 3pl.NOM(S)	kari NEG	dunga-y, go-NONPAST	kaba rain.ABS(S)	kada-nya-mun-ku. come-SUB-mun-DAT
	'They won't go in case it starts raining.'
(499)	Ngayu 1sg.NOM(A)	bayan house.ABS(O)	kida-l, sweep-NONPAST	dayirr-bunga-l, clean-CAUS-NONPAST
	kunka in case	ngayku 1sg.POSS.ABS(S)	jawun friend.ABS(S)	kada-nya-mun-ku. come-SUB-mun-DAT
	'I sweep the house and clean it in case my friend comes [to visit].'

Very often the subordinate clause is introduced by particle kunka 'in case', as in (499), or less commonly by nganganjirrka 'just in case'.

In the previous examples there are no common NPs, but co-referentiality of NPs is certainly possible. However it is not obligatory, nor indeed preferred, to delete any of them. Instead, pronoun reduction is commonly applied. Naturally, if co-referential NPs are retained in some form or other, co-referentiality need not be restricted. However, hypothetical reason constructions with common NP are few and far between and only the following possibilities have been attested:

A = S'

(500) Ngaji-ngka medicine wundi-l baral-ka, MF-ERG:pt(A) medicine.ABS(O) take-NONPAST road-DAT

kunka nyulu bambay-ma-nya-mun-ku. in.case 3sg.NOM(S) sick-INCHO-SUB-mun-DAT 'Grandfather takes medicine for the road in case he gets sick.'

A = LOC NP' (i.e. underlying A in passive)

(501) Maral-angka junkalu wundi-ny, kunka nyungu-ndu girl-ERG:pt(A) billycan.ABS(O) take-PAST in.case 3sg-LOC janbal nyaji-ji-nya-mun-ku.

quandong.ABS(S) see-ITR -SUB-mun-DAT

'The girl took a billycan in case she came across some quandong fruit. '

S = O'

(502) Kalnga-ngka juku-ji dunga-y, kunka kaya-ngka MB-ERG:pt(A) stick-cOMIT go-NONPAST in.case dog-ERG:pt(A)

nyunguny bayka-nya-mun-ku. 3sg.ACC(0) bite-SUB-mun-DAT 'Uncle goes with (takes) a stick in case the dog bites (attacks) him.'

Hypothetical reason clauses are regarded as somewhat cumbersome and are generally avoided in discourse, those with common NPs more so than those without. It is preferred to express the possibility of an event by an independent clause containing nguba 'perhaps' (see §3.9.2 (iii) and §4.9.3) which may in context be interpreted as the reason for another action, for example:

(503) N gayu marrkin wundi-l, nguba 1sg.NOM(A) gun.ABS(O) take-NONPAST perhaps 1sg-LOC bikibiki nyaJl-Jl-Y. pig(s).ABS(S) see-I1R-NONPAST '1 take a gun, perhaps (1) happen to see some pigs.' (ngayku-ndu)

4.4.5 Relative clauses

The definition of a relative clause depends on a formal or grammatical and a functional or semantic aspect. The formal aspect is that a relative clause is a constituent of a NP which may be indicated by case agreement,³⁸ and possible contiguity to the head noun. The functional aspect is that the relative clause supplies information about the head of this NP, thus helping to identify it. Both formal and functional criteria are important but, to my mind, the functional aspect is the more significant one. Thus a subordinate clause which is formally part of a NP, but is not relevant for the identification of this NP, cannot be termed a relative clause. Consider Comrie's (1981: 1 36) statement on the definition of a relative clause following a comparison between English and Turkish:

The lesson of this comparison is thus tha� we need a functional (semantic, cognitive) definition of relative clause, on the basis of which we can then proceed to compare relative clauses across languages, neglecting language-specific syntactic differences in our over-all definition of relative clause, but using them as a basis for our typology ...

There is no specific subordinating procedure in Kuku Yalanji which serves the sole purpose of relativising a NP according to the above definition. However, simultaneous subordinate constructions with different actor, as was briefly indicated in §4.4.4.1, may, with some caution, be interpreted as relative clauses. Consider (457H46 1) above and the following examples:

• (504) larramali-ngka bulki dalbarri-ny, manjal-ba thunderstorm-ERG:pt(A) cow.ABS(O) strike-PAST mountain-LOC (505) janjana-nya. stand.REDUP-SUB 'A thunderstorm struck the cow [which was/while it was] standing on the mountain.' Buliman-angka warru karrba-ny, police-ERG:pt(A) yg.man.ABS(O) grab-PAST house.ABS(O) janjarri-I-janjarri-nya. snoop-I-REDUP-SUB bayan 'The policeman apprehended the chap [who was/while he was] snooping
  • in the house.'

Such sentences are apparently open to two interpretations, termed "NP-relative" and ''Trelative" by K. Hale ( 1 976a). A NP-relative clause provides information about an argument in the main clause where main and subordinate clause share an identical argument, e.g. 'the cow which .. .' 'the chap who . . . '. A T-relative clause specifies the temporal setting of the

³⁸ This applies for instance in Latin and Greek; see Comrie (1981:1 46).

event in the main clause where two clauses make identical time reference, e.g. 'while (the cow) was standing', 'while (the chap) was snooping'.

On the evidence of these types of sentence it is impossible to decide whether the subordinate clause is embedded under the NP or adjoined under a lower S-node in deep structure. If it is true that both the NP-relative and the T-relative interpretation are possible, we can but speculate that their underlying structure is formally different, in which case there must be a transformation somewhere along the line which produces the same surface structure for both clause types.39 Alternatively, we can assume that their underlying formal structure is identical,⁴⁰ in which case a semantic rule would have to be responsible for giving the surface structure two different meanings. Thus positing a deep structure for these sentences is unsatisfactory either way. However, simultaneous subordinate structures with a perception verb in the main clause in Kuku Yalanji seem to suggest a structual difference between the NP-relative and the T-relative interpretation. Consider (462}-(465) and:

(506) a. Nyulu dingkar nyaji-l-nyaji-ny, bana 3sg.NOM(A) man.ABS(O) see-l-REDUP-PAST water.ABS(O) wulji-l-wulji-nya buwun-muny. bail.out-l-REDUP-SUB boat-ABL 'He watched the man [who was] bailing water from the boat. ' b. Nyulu nyaji-l-nyaji-ny, dingkar-angka bana 3sg.NOM(A) see-l-REDUP-PAST man-ERG:pt(A) water.ABS(O)

wulji-l-wulji-inya buwun-muny. bail.out-l-REDUP-SUB man-ERG:pt(A) 'He watched [whilelhow] the man was bailing water from the boat.'

In sentence (506b) it is fairly clear that the subordinate clause is derived from an adjoined clause: dingkarangka bana wuljilwuljiny buwunmuny 'the man was bailing water from the boat'. By comparison it would appear plausible to assume that in sentence (506a) we have a NP-relative clause dominated by the transitive object NP dingkar. But neither in sentences of type (504) and (505) nor in sentences of type (506) can we get away from the fact that the temporal interpretation is the overriding one: main and subordinate clause must refer to simultaneous events. A subordinate clause with -nyV cannot be used to express for instance 'the policeman apprehended the chap who had been snooping in the house yesterday' or 'a little while ago I saw the man who is now bailing out his boat'. Thus the characteristic identifying function of a relative clause is at best marginal in simultaneous verb + -nyV constructions, if it is present at all. To determine this, one would require the intuition of a native speaker, and it is quite possible that my own intuitions based on English and German would lead me to interpret, i.e. translate, a relative clause where this does not exist or is irrelevant for a native speaker of Kuku Yalanji.

Consider also verb + -nyV constructions with same actor, as for instance in (468)-(473) and:

(507)	Maral-angka dikal	nyaji-l-nyaji-l,	janjana-nya-mun-du.
	girl-ERG:pt(A) bird.ABS(O)	see- l -REDUP-NONPAST	stand.REDUP-SUB-mun-ERG
	'The girl watches the bird [while] standing up.'
	* 'The girl who is standing up watches the bird.'

³⁹ K. Hale (1976a:85) refers to this as the "extraction analysis".

⁴⁰ K. Hale (1 976a:85) refers to this as the "adjunction analysis".

The subordinate clause janjananyamundu (with deleted S), although part of the A NP as indicated by the case agreement, does not, according to native speakers, identify the girl but rather refers to the manner in which she performs the action of watching. On the other hand, some examples with co-referential actor again invite a T-relative interpretation in that the temporal co-occurrence of two actions is emphasised as in (469), (472) and (473). To summarise, the status of the most likely candidates for subordinate relative clauses in Kuku Yalanji, i.e. simultaneous subordinate constructions, is at least doubtfu1.41

To my knowledge, only independent clauses can unambiguously fulfil the function of further identifying the referent of an NP in Kuku Yalanji, as in:

• (508) Nyulu jalbu-ngku dunyu bawa-ny. 3sg.NOM(A) woman-ERG:pt(A) husband.ABS(O) leave-PAST Nyulu yinyay bundanda-y kangkal-ji. 3sg.NOM(S) there sit-REDUP-NONPAST own.child-cOMIT 'The woman left her husband. She [who] is sitting there with her child.'
• (509) Nyulu yinya mala-minya. Nyulu bama 3sg.NOM(S) that.ABS(S) expert-meat 3sg.NOM(S) Aborigine.ABS(S) wawubaja-nga dunga-y. river-LOC go-NONPAST 'That one is a good hunter. The Aborigine [who] is walking to the river.'
• (5 1 0) Ngayu ngawuya wawu daji-nka ngayku-wun-du I sg.NOM(A) turtle.ABS(O) want give-PURP I sg.POSS-LOC

jawun-undu. Nyulu yilayku kada-ny. friend-Loc:pt 3sg.NOM(S) yesterday come-PAST 'I want to give turtle [meat] to my friend. He [who] came yesterday.'

4. 4.6 Summary

Table 4.3 summarises the forms, functions and common NP deletion possibilities of subordinate clauses in Kuku Yalanji. As can be seen from this table there is a preponderance of obligatory co-referential S/A or co-referential 0 deletion. This indicates that in subordinate structures Kuku Yalanji apparently leans towards a S/A pivot which disinguishes between the grammatical function of S/A as subject on the one hand and 0 as transitive object on the other hand. Thus, in spite of its ergative/absolutive case-marking system, Kuku Yalanji displays to some degree characteristics of a nominative/accusative language in subordinate constructions.

Complex sentences with more than one subordinate clause are not commonly used in Kuku Yalanji, although they are possible. Of all subordinate clauses the purposive clause is most likely to be followed by another clause that has some logical connection, for example:

41 Similar, although in some respects slightly different, problems in determining relative clauses exist in Yidiny (Dixon 1 977 :423-428).

184 Chapter 4

Function	Verb form	Main NP	Common NPs42
Purposive	stem + -nkV	S, A O oblique	Sø, Aø, (O), (oblique) Oø, (S), (A), (oblique) (S), (A), (O), (oblique)
Precautionary	stem + -nyji	S, A O oblique	Sø, Aø, (O) Oø, (S), (A) (S), (A) or no common NP
Successive	stem + -nyjiku	S, A O	Sø, Aø, (O) Oø, (S), (A) or no common NP
Simultaneous action	(reduplicated) stem + -nyV + ABS/ERG (reduplicated) stem + -nyV	S, A O	Sø, Aø Sø, Aø
with perception verb	(reduplicated) stem + -nyV (+ ABS?) (reduplicated) stem + -nyV (+ oblique case)	Oø oblique	S, A Sø, Aø
Causal	stem + -nyV	S A O	Sø, Aø (S), (A) (S), (A), (O)
Prior event	wawu + stem + -nyV + ABL	S O	Sø Sø, Aø
	verb not subordinate (wawu + O NP + ABL)	S	Aø
Hypothetical reason	stem + -nyV + DAT	A S	S O (others?) or no common NP

Table 4.3: Functions and forms of subordinate clauses in Kuku Yalanji

⁴² NP $\emptyset$ = obligatory deletion.

(NP) = optional deletion or pronoun reduction.

(511)	Ngayu 1sg.NOM(S)	wawubaja-nga river-LOC	dunga-y go-NONPAST	kuyu fish.ABS(O)	mani-nka ... get-PURP
'I go to the river to catch a fish...'

a.	... dakuy-ma-nyji.	'... lest I get hungry.'
b.	... dakuy-ma-nya-mun-ku.	'... in case I get hungry.'
c.	... bayan kida-nyjiku.	'... before I clean the house.'
d.	... wawu-bayan-muny kiday.	'... after I clean the house.'

4.5 Possession

Like most Australian languages Kuku Yalanji has the means to distinguish syntactically between alienable and inalienable possession. Alienable possession includes concrete and abstract things, e.g. artefacts, animals, language, and kinship relations. The possessor is usually animate and normally human. However, the construction used for alienable possession may also be used for inalienable possession, i.e. part-whole relationship, and where the possessor is inanimate; it will therefore be termed 'general (alienable) possession'. On the other hand, different constructions are available to express part-whole relationship or inanimate possessor, which cannot be used for alienable possession by an animate possessor.

4.5. 1 General (alienable) possession

General (alienable) possession is marked by the stem-forming suffixes:

-(V)mu (Yalanji), -(V)ngV (Nyungkul) / -zero inflection

-(V)ndVmun- / -non-zero inflection (see §3.2.3.2)

on all words of the possessive NP, which modifies the head noun. Thus an alienable possessive NP has the structure:

[[NP] + case [NP + POSs] + case] NP

A possessive stem always takes neutral inflection. The head noun and the POSS NP may occur in any order. Some examples of a possessive NP in absolutive case, showing both the Yalanji and Nyungkul suffixes, are:

• (5 1 2) Nganjan -amu kaya jinbal-ku wanarri-y. -anga father-POSS.ABS(S) dog.ABS(S) fast-ADV run.REDUP-NONPAST 'Father's dog is running fast. '
• (5 1 3) Kuyu-mu jindi jabajaba. -ngu fish-POSS.ABS(S) scale.ABS(S) big/plentiful 'The scales of the fish are big and plentiful.'
• (5 1 4) Kuyu ngulkurr wawubaja-mu. -nga fish.ABS(S) good river-POSSABS(S) 'The fish of in (lit. of) the river are good.'

1 86 Chapter 4

Example (5 1 3) expresses inalienable possession and in (5 1 4) the 'possessor' is inanimate. (Note that in the Nyungkul dialect the possessive suffix cannot be distinguished from neutral locative inflection if the noun ends in a vowel.)

Some examples of possessive NPs in other than absolutive cases are:

• (515) Dingkar-andamun-du yaba-ngka bikibiki kuni-ny. man-POSS-ERG(A) eB-ERG:pt(A) pig.ABS(O) kill-PAST 'The man's elder brother killed a pig. '
• (5 1 6) N gayu binal-ma-l bama-ndamun-ku 1sg.NOM(S) know-INCHO-NONPAST Aborigine-POSS-DAT food-DAT 'I am learning about Aborigines' food.' mayi-ka.
• (5 1 7) Nganjin yambayamba Toyota-nga mumba-ny 3pl.exc.NOM(A) everything.ABS(O) Toyota-LOC put.in-PAST

Chris-andamun-bu. Chris-POSS-LOC 'We put everything into Chris's Toyota.'

Example (5 1 5) illustrates kinship relation expressed by alienable possession.

Very seldom does a possessive NP contain a 'possessor of a possessor'. Such constructions contain a regular possessive form without a second possessive suffix and optional case inflection, for example:

(5 1 8) Ngayu wawu-kari yunu-wun-ku nganjan-andamun-ku kaya-nka. I sg.NOM(S) like-PRIV 2sg-POSS-DAT father-POSS-DAT dog-DAT:pt 'I don't like your father's dog. '

However, there seems to be a strong tendency to use the POSS + LOC form for 'possessor of' possessor', no matter what case the head noun is in. In (5 1 8) yunu-wun-bu (LOC) is just as acceptable, and indeed preferred over the dative form.

4.5.2 Inalienable possession

Inalienable possession may be expressed by a general possession NP as in (5 1 3), and:

(5 1 9) Nyungu mungka duna. 3sg.POSS.ABS(S) hair.ABS(S) wet 'His hair is wet.'

But it is more common to place whole and part in apposition within the NP, both agreeing in case, but without a possessive suffix at all. Such an NP has the form:

[whole + case part + case]NP

Examples of this can be seen for instance in (483) and (486) and also:

(520)	Nyulu	kuyu	biji	kayi-ny.
	3sg.NOM(A)	fish.ABS(O)	tail.ABS(O)	hook-PAST
	'He hooked the fish (on) the tail.'

(52 1) Nyulu kulji dara-y-mani-ny maral-anda jina-nga. 3sg.NOM(A) stone.ABS(O) fall-y-CAUS-PAST girl-LOC:pt foot-LOC 'He dropped the stone on the girl's foot.'

As shown previously in (5 1 5) kinship relations are expressed by general alienable possession. However, when talking about someone's offspring, e.g. own child or grandchild, this may also be indicated by inalienable possession:

• (522) Ngayu kangkal jambul. I sg.NOM(S) own.child.ABS(S) two 'I have two children.'
• (523) Ngayu kaminjarr jalbu-jalbu wubul. I sg.NOM(S) gr.child.ABS(S) woman-REDUP.NOM(S) many 'I have many granddaughters.' (Kaminjarr refers to female ego's daughter's child and male ego's son's child.)

The speaker in both instances was a female, so by a little stretch of imagination one could interpret these constructions as whole-part relationship. (I do not know whether a male could say the same thing.)

4.5.3 Inanimate possessor

Again, an inanimate possessor may have the general possessive form as in (5 1 4), but this is somewhat unusual. More common is the use of locative or ablative, indicating the location or origin of something. For instance:

(524)	Birra	juku-ngu/juku-muny	yalbay-bajaku.
	leaves.ABS(S)	tree-LOC/tree-ABL	big-very
	'The leaves on this tree/from this tree are very big.'

The locative refers to the leaves that are actually on the tree, whereas the ablative implies that the leaves have been plucked or fallen off. Similarly:

(525) Mayi yinya-nga/-muny juku-ngu/-muny ngulkurr nuka-ji-y fruit.ABS(S) that-LOO-ABL tree-LOOABL good eat-ITR-NONPAST 'Fruit onlfrom that tree is good to eat.'

On the other hand, ablative was not accepted in the following sentence:

(526)	Dumbul	marra-nga/-mu/*-muny	dandi.
	shell.ABS(S)	zamia.nut-LOC/-POSS.ABS(S)/*-ABL 'The shell on/of/*from the zamia nut is hard.'	hard

(Maybe once the shells are peeled off, they are all crumbled up and softened.) Note that the above examples are in Yalanji dialect, which clearly distinguishes in form between locative and possessive. Therefore juku-ngu and marra-nga are definitely in locative case.

4.5.4 Former possession

To indicate a former possession, the particle (or suffix?) mali is employed following a noun or pronoun in locative or ablative case. The choice of case differs between speakers, some preferring locative, others ablative, for example:

(527) Yinya ngayku-ndu/-ndumuny(-)mali dunyu. that 1sg-LOC/-ABL(-)former.possession husband 'That is my former husband.'

1 88 Chapter 4

(528) Yinya bayan jalbu-ndul-ndumuny(-)mali. that house woman-Loci-ABL(-)former.possession 'That house formerly belonged to a woman.'

At this stage it is not clear whether mali is a particle or a suffix or a clitic.43

4.5.5 Relationship between possessive and locative

In some Australian languages, for instance Guugu Yimidhirr (Haviland 1 979a) and Gumbaynggir (Eades 1 979), there is a close relationship between possessive (or genitive) and dative which marks the recipient of an act of giving. A similar close relationship pertains in Kuku Yalanji between possessive and locative.

In Kuku Yalanji the recipient of an act of giving may be in locative case as was shown in the discussion of case roles (§4.1 .4.4 (iii» , as for instance:

(529) Bula jalbu-bulal-angka mayi kubarr 3du.NOM(A) woman-pair-ERG:pt(A) food.ABS(O) eel.ABS(O) daji-ny dingkar-anda. give-PAsT man-LOC:pt 'The two women gave the eel to the man.'

But the recipient may also be a possessive NP in 0 function:

(530)	Ngayu	ngamu-mu	mayi	daji-ny.
	1sg.NOM(A)	mother-POSS.ABS(O)	food.ABS(O)	give-PAST
	'I gave mother [her] food.' (lit. I gave mother's food)

It may be arguable whether the whole possessive NP in constructions of the above type is in 0 function or only the 'possessed' part. If the possessive noun, i.e. ngamu-mu, were not in o function, possessive in connection with daja-l 'give' would have to be interpreted as an inflection indicating 'goal' rather than a derivation. However, possessive apparently cannot be used if the thing that is given is omitted, e.g. 'I gave (it) to mother'. In such instances only locative is possible. Furthermore, possessive cannot be substituted for locative to indicate 'goal' in other three argument verbs such as milbi-l 'show'. I therefore prefer the interpretation of possessive as a derivation only and the whole possessive NP being in 0 function in connection with 'give'.

There may be a slight difference in the interpretation of the above sentences in that a possessive NP in 0 function implies that the recipient has a rightful entitlement to the thing given, whereas a locative form indicates spontaneous giving without obligation. However, this is mainly conjecture on my part based on the Aboriginal tradition of obligatory sharing among kin and on somewhat similar constructions in English and other languages such as 'I gave the boy his breakfast' versus 'I gave the clothes to the Smith Family'. But the fact remains that the recipient of goods may be either a possessive or a locative NP.

The fact that an inanimate possessor may involve locative (or ablative) is probably of no great value for this argument, since usually there is a real location involved, such as 'leaves on the tree'. However, it is noteworthy that possessive may be used as well to express this concept; and note the use of a noun or pronoun in locative (or ablative) case in conjunction

⁴³ Compare "general genitive inflection" -mi in DyirbaJ which indicates a "past owner" in contrast to "simple genitive" -nu which expresses "present possession" (Dixon 1 972: 1 08).

with (-)mali to express former possession. Consider also the morphology of the possessive stem-forming suffix -(V)ndVmun-, which is used if a non-zero inflection follows. This suffix looks remarkably like potent locative inflection plus catalytic suffix: -(V)ndV- + -mun- . But this similarity does not warrant the general reinterpretation of locative as a derivation, since this form occurs only in possessive function.

More circumstantial evidence for some relationship between locative and possessive can be gleaned from the fact that a 'possessor of a possessor' is preferred to have the form 'possessive pronoun stem + LOC' regardless of the case marking of the NP to which it belongs. Finally, there is the matter of the set of kinship terms which allow either potent or neutral case inflection. In a possessive NP these may take either the regular possessive suffix or a suffix which is identical to the neutral locative inflection (see §3.2.4).

For the time being, the conspicuous resemblance between possessive and locative in Kuku Yalanji provides food for thought, rather than giving any conclusive evidence as to the exact nature of the relationship between the two.44

4.6 Comitative and privative

Comitative and privative forms of nouns are opposite in meaning, which is basically 'with Noun' and 'without Noun' respectively, but they both serve as modifiers and as such may be incorporated into a wider NP which has the structure:

$$\begin{array}{rcl} \text{[[NP1] + case]} & \text{[NP2 + COMIT] + case} \text{[NP]}\ & \text{[}_{\text{ass}}\text{]} \text{[}_{\text{ann}}\text{]} & \text{[}_{\text{PRIV}}\text{]} \end{array}$$

NP1 may be a personal or demonstrative pronoun or a single noun or both for either construction. NP2 may have the same constituents as NP1 or a noun plus an adjective, both with comitative suffix, in a comitative noun phrase. A privative construction, on the other hand, can have only a single noun or adjective in NP2. Like an adjective, a comitative or privative form on its own may function as predicator in an equational sentence.

4. 6. 1 Comitative

A noun with comitative suffix -(i)ji (,neutral'), -(V)nji ('potent'), -(i)jirr- (see §3.2.3.3) may have various functions in a clause. Following the list set out in Dixon, ed. (1 976:306- 307), with some amendments that apply for Kuku Yalanji, these are:

(i) Attribute

In this function the comitative noun resembles an adjective and may be used as predicator in an equational sentence or as a modifier within a wider NP, which will be discussed later. Note that although translations of some of the sentences below necessarily include 'have' the comitative does not indicate a relationship of possession, but rather a characteristic. The attributive functions of comitative can be classed under the following headings:

⁴⁴ Note that in some languages, e.g. Russian, the way to say 'I have X' is 'X is at me' using locative case or an equivalent preposition. See Clark (1 978) and Lyons (1977:722-723) for a general discussion on the relation between location and possession.

190 Chapter 4

• (a) mental or corporeal state of person or animal:
  • (531) Yinya kaya kuli-ji. that.ABs(s) dog.ABS(S) rage-COMIT 'That dog is vicious. '
  • (532) Nyulu karrkay kumbu-ji. 3sg.NOM(S) child.ABS(S) urine-COMIT 'The child needs to have a pee.'
• (b) physical characteristic of someone or something:
  • (533) falun mulngku-ji. sea.ABS(S) wave-COMIT 'The sea is rough.'
  • (534) Diburr YlJzrr-ji kima-ji. egg.ABS(S) shell-COMIT soft-COMIT 'The egg has a soft shell.'
  • (535) Yinya dingkar walarr-iji. that.ABS(S) man.ABS(S) beard-COMIT 'That man has a beard.'

(c) locational characteristic of people:

(536) Jana bama ngalkal-ji. 3pl.NOM(S) Aborigine.ABS(S?) lightJorest-COMIT 'They are people of light forest country.' or 'The people are of the light forest country.'

Bama in the last example may be either part of the S NP or the head noun of the comitative NP according to intonation pattern. The same applies for the following two sentences.

(d) alienable possession as characteristic:

(537)	Nyulu	jalbu	kambi-ji	wubul-ji.
	3sg.NOM(S)	woman.ABS(S?)	clothes-COMIT	many-COMIT
	'The woman has	many clothes.'	(lit. is	a many-clothes-owner)

(538) Yinya dingkar kalka-ji kulur-ji. that.ABS(S) man.ABS(S?) spear-COMIT three-COMIT 'That man has three spears.' (lit. is a three-spear owner)

(e) In a similar attributive function comitative may be used to describe the two main seasons:

(539) wungar-iji sun-COMIT 'hot season'

(540) bulur-iji early.morning.chillldew-coMIT 'cold season'

(541) Nyungu ngaji wungar-iji wula-ny. 3sg.POSS.ABS(S) MF.ABS(S) sun-COMIT die-PAST 'His grandfather died in the hot season.'

Some examples of NPs containing comitative attributes in various cases are:

• (542) Warru-ngku jarba kuli-ji kuni-ny. yg.man-ERG:pt(A) snake.ABS(O) rage-COMIT.ABS(O) kill-PAST 'The young man killed the vicious snake. '
• (543) walarr-iji-ngka ngalkun wubul Dingkar-angka man-ERG:pt(A) beard-COMIT -ERG:pt(A) mullet.ABS(O) many.ABS(O) dama-ny. spear-PAST 'The man with the beard speared many mullet.'
• (544) Ngayu wawu jirray juku-ku birra-jirr-ka I sg.NOM(S) like much tree-DAT leaf-COMIT-DAT

jabajaba-jirr-ka. bigiplentiful-cOMIT-DAT 'I like very much the tree with the many big leaves.'

(545) Babi-ngka yangka daji-ny karrkay-anda FM-ERG:pt(A) green.ant.ABS(O) give-PAST child-LOc:pt ngiki-ji-nda. cough-COMIT-LOc:pt 'Grandmother gave green ant [juice] to the child with the cough.'

• (546) Nyulu kada-ny bayan-muny yinya-muny juku-jirr-muny. 3sg.NOM(S) come-PAST house-ABL that-ABL tree-COMIT-ABL 'He came from that house with the tree.'
• (ii) Accompaniment:
• (a) A person in motion or at rest with animate company:
  • (547) Ngayu bimay-anji kada-ny. I sg.NOM(S) FZ-COMIT:pt come-PAST 'I came with [my] aunt. '
  • (548) Warru-warru bikibiki-nga dunga-y kaya-nji. yg.man-REDUP.ABS(S) pig-LOC go-NONPAST dog-coMIT:pt 'The young men are going out for pigs with a dog/dogs.'
  • (549) Nyulu bayan-ba bundanda-y ngawa-nji. 3sg.NOM(S) house-LOC sit.REDUP-NONPAST baby-coMIT:pt 'She is sitting in the house with the baby.'
• (b) A person in motion with helpful or non-helpful inanimate implement:
  • (550) Nyulu dingki-ji badabada dunga-ny wawubaja-nga. 3sg.NOM(S) dinghy-COMIT down go-PAST river-LOC 'He went down the river with a dinghy. '

1 92 Chapter 4

(55 1) Wulman dunga-y kalka-ji waymbil-ji. old.man.ABS(S) go-NONPAST spear-COMIT axe-COMIT 'The old man is going out with a spear and an axe.'

If, on the other hand, a body part is involved in the motion, this must be in instrumental case; see (245) and (246) in §4. 1 .4.4 (i).

(c) A person at rest with inanimate thing:

(552) Jawun bundanda-y friend.ABS(S) sit.REDUP-NONP AST wulngku-ji. tape.recorder-COMIT kaban-kaban-ji paper/book-REDUP-COMIT '[Our] friend is sitting with papers/books and a tape-recorder.' (wulngku originally: 'song, tune')

(d) An animate being performing an action (other than movement) with an implement or together with another animate being. In Kuku Yalanji this use of comitative is apparently restricted to intransitive sentences and therefore differs from the instrumental use of comitative with transitive verbs attested for other languages (see Dixon, ed. 1 976:306).

(553)	Yinya	kaya	kinka-n-kinka-y	bajibay-ji.
			that.ABS(S) dog.ABS(S) play-n-REDUP-NONPAST bone-COMIT
		'That dog is playing with a bone.'

(554) Warru wuri-n-wuri-y jinkurr-ji. yg.man.ABS(S) dance-n-REDUP-NONPAST yZ-COMIT 'The young man is dancing with his younger sister. '

It will have been noticed that the examples of comitative in 'accompaniment' function do not indicate absolutive case agreement with the head noun. In fact, because an 'accompaniment' interpretation of comitative is possible only in intransitive sentences it cannot be ascertained whether the comitative form is part of the S NP since absolutive has zero realisation. In the absence of definite evidence for case agreement one could argue that comitative in 'accompaniment' sense has a more inflectional characteristic, whereas it is clearly a derivation in attributive function. Note also that an 'accompaniment' comitative form is preferred in clause-final position, whereas an attributive comitative form is normally found adjacent to the head noun. However, this aspect is obviously not sufficient support for an interpretation of 'accompaniment' comitative as inflection. I therefore regard comitative as having only one grammatical function, i.e. that of derivational suffix.

4. 6.2 Privative

Privative forms (see §3 .2.3.4) are used less frequently than comitative forms in Kuku Yalanji. Their attributive functions as outlined below correspond largely to those of comitative, but privative forms are preferred as predicator in equational sentences and less favoured as part of a wider NP.

(i) Privative forms in adjectival or attributive function:

(a) Mental or corporeal state (cf. comitative (i)(a))

• (555) Jana kiru-kari. 3pl.NOM(S) brain-PRIV 'They are stupid.'
• (556) Ngayu ngiki-kari, ngulkurr-ma-ny. I sg.NOM(S) coughlcold-PRIV good-INCHO-PAST 'J don't have a cold, [I am] well again.'
• (b) Physical characteristic (cf. comitative (i)(b))
  • (557) Nyulu wulbuman dirra-kari. 3sg.NOM(S) old.woman.ABS(S) tooth-PRIV 'The old woman has no teeth.'
  • (558) Nyulu mungka-kari-ngka ngayku dambal 3sg.NOM(A) hair-PRIV-ERG:pt(A) I sg.POSS.ABS(O) shoes.ABS(O) ngaki-ny. steal-PAST 'The one without hair stole my shoes.'
• (c) 'Non-possession' (cf. comitative (i)(d))
  • (559) Nyulu bayan-kari jibabadi-y. 3sg.NOM(S) house-PRIV.ABS(S) be.sad-NONPAST 'The homeless one is sad. '
  • (560) Nganjin tealeaf-kari. I sg.exc.NOM(S) tea-PRIV 'We have no tea.'
  • (56 1) Ngayu kambi daji-ny jalbu-ndu kangkal-kari-nda I sg.NOM(A) clothes.ABS(O) give-PAST woman-LOc:pt own.child-PRIV-LOC 'J gave clothes to the childless woman.'

'Non-accompaniment' use (cf. comitative (ii)) of a privative form is rare. Some examples are:

• (562) Nyulu dunga-ny marrkin-kari. 3sg.NOM(S) gO-PAST rifle-PRIV 'He went without a rifle.'
• (563) Ngawa warngku-wuna-y kararr-kari. baby.ABS(S) sleep-lie-PAST blanket-PRIV 'The baby sleeps without a blanket.'

For motion or rest without animate accompaniment (cf. comitative (ii)(a)) the privative form is not favoured at all but a paraphrase like the following is preferred:

(564) Ngayu kari jilba-dunga-y jalbu-nji, I sg.NOM(S) NEG walk-go-NONPAST woman-COMIT ngayu-rrku dunga-y. I sg.NOM(S)-EMPH4 go-NONPAST 'J don't go walkabout with the women, J go by myself. '

1 94 Chapter 4

4.7 Imperatives

An imperative can be formed from any verbal root or stem in Kuku Yalanji (see §3.8.4.2 (iii)). Normally, it has an explicit or implicit second person pronoun in S or A function. If mentioned, the subject is strongly preferred in initial position, but it may also occur elsewhere in the sentence:

• (565) (Yurra) julurri-ji-ka! 2pl.NOM(S) wash-REF-IMP 'Wash yourselves !'
• (566) (Yundu) ngamu 2sg.NOM(A) mother.ABS(O) 'Don't blame [your] mother !' kari nyuya! NEG blame.lMP
• (567) (Yundu) yinya dara-y-mana! 2sg.NOM(A) that.ABS(O) fall-y-CAUS.lMP 'Drop that!'

If the subject NP is included, the imperative can be distinguished from a statement by the intonation contour (see §2.6.3).

A reduplicated verb stem would prompt the addressee to continue with some activity, as typically in:

(568) (Yubal) bundanda-ka, ngayu kada-y-baja! 2du.NOM(S) sit.REDUP-IMP I sg.NOM(S) come-NONPAST-again '(Y ou two) keep sitting there, I will come back.'

Occasionally, a first person inclusive non-singular pronoun may be the subject of an imperative clause. This expresses an encouragement to do something, as for instance:

(569) (N gana) kada-ri-ka-da! I pI. inc.NOM(S) come-PLS-IMP-TEMPI "Come on, let's go now!"

An imperative sentence may contain the particle balu 'desist' (see §4.9.2 (ii)) , which implies that the addressee should leave someone or something alone. This has only been attested with intransitive imperative sentences:

(570)	Balu dungay!	'Let him/her go!'
(571)	Balu wajujika!	'Let it burn!'

But there is no apparent reason why balu should not be used with a transitive imperative sentence as well, for example:

(572) ? Balu nyulu mayi nuka! 'Let him eat his food!'

The temporal verbal clitic -da, which means 'following with respect to a preceding action' (see §3.10.1 ), is very often used with imperatives. While a plain imperative form such as mayi nuka 'eat food!' may be regarded as a general prompt of the type 'eat something! (you must be hungry)' or 'go on, eat your food, don't mind me!', the form mayi nuka-da urges the addressee to eat now and to discontinue any other activity. A stronger command for immediate action can be expressed by the clitic -(ng)Vrr 'straight away/first of all' (see §3.1 0.1), for example:

(573) Nuka-ngarr! OR Mayi-ngarr nuka! 'Eat straight away (before you do anything else)!'

The urgency in -(ng)Vrr which forbids any other possible activity by the addressee was illustrated in the following way. My informant Louisa Smith wanted her daughter to come and help with her sick father. She called: "Nora kaday, hand daya!" (sic) "Nora come here and give me a hand!" When this brought no response she called again: "Nora kadayarr! " When still no Nora arrived, Louisa exclaimed forcefully: "Nora, come here first!" (which is in fact a very common strong English imperative used by the Kuku Yalanji people).

Apparently there is no politeness distinction between imperatives with different clitics. As a very well-mannered person assured me, she could use all imperative forms when speaking to her children as well as when speaking to the pastor. R. Hershberger (l 964b:52) distinguishes between -ngarr as polite imperative marker and -rr as emphatic imperative marker. According to my data these are allomorphs of the same form (see §3. 10.1) and, as indicated above, do not imply politeness/emphatic distinction.

4.8 Questions

Polar questions, requiring a yes/no answer, are marked only by intonation contour (see §2.6.3) and para-linguistic features, such as facial expression, in Kuku Yalanji.

(574)	Yundu	ngayku	kangkal	nyaji-ny?
	2sg.NOM(A)	1sg.POSS.ABS(O)	own.child.ABS(O)	see-PAST
		'Have you seen my child?'

(575) Nyulu kada-ny-baja? 3sg.NOM(S) come-PAsT-again 'Did he come back?'

For non-polar questions there is a range of interrogative pronouns, which were discussed in §3.6.1, including some examples. Further examples of the use of interrogative pronouns are:

• (576) Wanya kada-ny maja-nji? who.ABS(S) come-PAST bOSS-COMIT:pt 'Who came with the boss?'
• (577) Wanya-ngka ngayku dambal wundi-ny? who-ERG:pt(A) I sg.POSS.ABS(O) shoes.ABS(O) take-PAST 'Who took my shoes?'
• (578) Yundu wanya-nji Cooktown dunga-ny? 2sg.NOM(S) who-COMIT:pt Cooktown gO-PAST 'With whom did you go to Cooktown?'
• (579) Yundu wanyu mumba-ny balji-nga? 2sg.NOM(A) what.ABS(O) put.in-PAST bag-LOC 'What did you put in the bag?'
• (580) Wanyurri-ngku mayi-ka yundu wawu? what-DAT food-DAT 2sg.NOM(S) want/like 'What (sort of) food do you wantllike?
• (58 1) Nyulu wanja-muny? 3sg.NOM(S) where-ABL 'Where is he from?'
• (582) Wanjawanja yundu kada-y-baja, wungariji? when 2sg.NOM(S) come-NONPAST-again hot.season 'When will you come back, (in the) hot season?
• (583) Wanjarrinya bama yinyay how.many Aborigine.ABS(S) there 'How many Aborigines live there?' bundanda-y? sit.REDUP-NONP AST

4.9 Particles

In this section we will look at particles with a function at clause or sentence level, i.e. adjectival, comparison, negation and discourse particles. (For morphological features of these particles see §3.9.)

4.9. 1 Adjectival particles

Non-inflecting adjectival particles (see §3.9.1) function as predicators, taking as complement either a noun in dative case or a purposive clause. The following sentences exemplify the use of all attested adjectival particles:

• (584) Jana nyiku binal-kari kuku-ku. 3pl.NOM(S) today know-PRIV language-DAT 'They don't know the language nowadays.'
• (585) Jalbu binal balji wukurri-nka. woman.ABS(S) know dillybag.ABS(O) weave-PURP 'The woman knows how to weave dillybags.'
• (586) Nyulu dingkar juburr bulki-ka. 3sg.NOM(S) man.ABS(S) be.good.at cattle-DAT 'The man is good at/with cattle. '
• (587) Ngayu yinil bilngkumu-nku. I sg.NOM(S) afraid crocodile-DAT 'I'm afraid of crocodiles.'
• (588) Karrkay-karrkay wawu jirray kinka-n-kinka-nka. child-REDUP.ABS(S) like much play-n-REDUP-PURP 'The children love to play. '
• (589) Nyulu wawu kalka balka-nka. 3sg.NOM(S) want spear.ABS(O) make-PURP 'He wants to make a spear.'
• (590) Karrkay wawu ngaki-ji-nka. child.ABS(S) wantllike hide-REF-PURP 'The child wantsllikes to hide. '

As can be seen in (585) and (589) the subject NP does not change to y function if the following purposive clause contains a transitive verb. This shows that the clause with adjectival particle must be regarded as an intransitive main clause and the adjoined purposive clause as subordinate. In Guugu Yimidhirr (Ra viland 1 979a: 1 54) a distinction between main and subordinate clause with similar adjectival particles is not clear cut, since the subject NP may be in A function if the purposive complement clause contains a transitive verb.

In derived inchoative verb form, adjectival particles maintain the same syntactic type of complement, for example:

(59 1) Ngayu binalmal kukuku. 'I learn language. '

In derived causative verb form, however, they naturally are accompanied by a noun in 0 function, with an optional goal as dative noun or purposive clause, for example:

(592)	Kambakamba-ngka	nganya	binal-bunga-ny	mayi-ka
	old.women-ERG:pt(A)	1sg.ACC(O)	know-CAUS-PAST	food-DAT
	jarruka	baka-nka.
	scrub.hen.ABS(O)	dig-PURP
	'The old women taught me about food (and) how to dig scrub hen (eggs).'

4.9.2 Negation particles

The two negation particles kari 'not' (NEG) and balu 'desist, give up' (NEG2), see §3.9.2 (ii), have different application possibilities according to their semantic content.

(i) Particle kari (see also function as privative suffix in §4.6.2 and discourse particle 'but, however' in §4.3.2 and §5.5.3) may negate various clause constituents depending on word order and stress. When negating the predicate, kari bears primary clause stress and immediately precedes the verb or the 0 NP:

(593) Ngayu jungkalu k'ari dumbarri-ny. I sg.NOM(A) billycan.ABS(O) NEG break-PAST

or:

N gayu k'ari jungkalu dumbarriny. 'I did not break the billycan.'

If a noun in subject or object function is to be negated, kari is preferred to follow this noun, but it is the noun which bears primary stress:

• (594) Ngayu j'ungkalu kari dumbarriny. 'I did not break the billycan [but something else].'
• (595) N'gayu kari jungkalu dumbarriny. I did not break the billycan [but something else did].'
• (596) D'ubu kari kankaday, dingkar. Not a ghost is coming, [but] a man.'

Compare (596) to (597), where kari is stressed:

1 98 Chapter 4

(597) Dubu kari k'ankaday, dungandungay. 'The ghost is not coming, [but] going. '

If an adjective is to be negated, kari may precede this, in which case kari is stressed, or it may follow the adjective, in which case the adjective is stressed:

(598) a. Ngayu k'ari dakuy. 'I am not hungry'

b. Ngayu d'akuy(-)kari.

In (598b) kari can probably be interpreted as privative suffix which means that the adjective + -kari as a whole has predicator function. The same interpretation of kari as privative also applies with adjectival particles which are invariably followed by it (see §3.9. 1 .).

(ii) Particle balu has a somewhat ambiguous grammatical status. Its syntactic behaviour is like that of adjectival particles in that it governs a noun in dative case or a purposive clause, and the subject NP in a balu sentence is always in S function (see syntax of adjectival particles in §4.9.1). On the other hand, balu does not have the morphological possibilities of adjectival particles, i.e. verbal stem-formation, privative, and combination with -bajaku 'very'. (For morphology of adjectival particles see §3.9.1 .) Because of these morphological differences I do not include balu among the adjectival particles.

The negation function of balu as 'desist, give up, refrain' is illustrated in the following examples:

(599)	Nyulu	warru	balu	dama-nka	ngawuya.
	3sg.NOM(S)	yg.man.ABS(S)	NEG2	spear-PURP	turtle.ABS(O)
	"The young man could not spear the turtle." (i.e. he made an unsuccessful
	attempt at spearing it and gave up)

• (600) N gayu balu kuniwa-nka. I sg.NOM(S) NEG2 fight-PURP "I don't like fighting." (i.e. I disapprove of fighting and refrain from doing it myself)
• (601) Ngayu balu buymbi-nka. I sg.NOM(S) NEG2 sucklsmoke.tobacco-PURP "I don't like to smoke." (i.e. I know it is bad for my health and therefore abstain from it)
• (602) N gayu balu mayngku-ku wawumu-ku. I sg.NOM(S) NEG2 mango-DAT half.ripe-DAT "I don't like half-ripe mangoes." (i.e. they give me indigestion and I therefore refrain from eating them)

The contrast between balu and kari is that kari simply negates whereas balu implies a certain reason, i.e. inability or disapproval, why something is not done. Balu also differs from wawu-kari 'don't want, don't like' in that it indicates disapproval as a general principle whereas wawu-kari has more specific reference. Compare (601) with the following:

(603) Ngayu wawukari buymbi-nka. 'I don't want/don't like to smoke.' (perhaps because I have a cold, but normally I do smoke)

Consider also the 'collateral' discourse function of balu (see §5.5.3) as indicating a false assumption, i.e. 'give up a preconceived idea (because of contrary evidence)', and balu in imperatives as 'leave it be, desist from what you are doing' (see §4.7).

4.9.3 Other particles

The use of co-ordination particles kari and yamba 'but, however', kunka and nganganjirrka 'in case', and kaki 'if, when' has been illustrated in §4.3.2. The two remaining particles that need further consideration are comparison particle yala, and nguba 'perhaps, maybe'.

(i) Comparison particle yala '(be) like, (be) the same' (COMP) (see §3.9.2 (i» , is followed by an NP in absolutive case, or nominative case if a pronoun, for example:

(604)	Nyulu	yala	nyungu	nganjan.
	3sg.NOM(S)	COMP	3sg.POSS.ABS	father.ABS
	'He is (just) like his father.'

• (605) Eskimo mala-minya yala bama. eskimo.ABs(S) expert-meat COMP Aborigine.ABS 'Eskimos are expert hunters like Aborigines.'
• (606) Ngayu wawu jirray mayi-ka yala mayngku. I sg.NOM(S) like much fruit-DAT same mango-ABS 'I like very much fruit [that is] like a mango.'

Note that in other comparison constructions that do not contain particle yala the compared component takes dative case; see §4. 1 .4.4 (ii).

R. Hershberger (1 979:82-83) attested yala in an additional function as co-ordination particle, for example:

(607) N gayu mayi wundi-ny yala nyulu ngadiku I sg.NOM(A) food.ABS(O) bring-PAsT COMP 3sg.NOM(A) long. ago wundi-ny. bring-PAST 'I brought food like he brought it long ago.' (quoted from Hershberger; my spelling and glosses)

But sentences of this type were not whole-heartedly accepted by my informants. Because of the insufficient evidence I have not included yala among the co-ordination particles.

(ii) Particle nguba 'perhaps, maybe' indicates that the speaker makes an assumption which may or may not hold true, or that he presents possible alternatives, for example:

• (608) Jana wanjabu? N guba jana ngaki-ji-y. 3pl.NOM(S) where maybe 3pl.NOM(S) hide-REF-NONPAST 'Where are they? Maybe they are hiding.'
• (609) Nguba ngayu nyiku kada-y-baja, nguba wunkuny. maybe I sg.NOM(S) today cOme-NONPAST-again maybe tomorrow 'Maybe I come back today (or) maybe tomorrow. '

200 Chapter 4

(6 1 0) N guba ngayu kuyu nyama-l, nguba dinda-l. maybe I sg.NOM(A) fish.ABs(o) stew-NONPASf 'Maybe I 'll stew the fish, maybe [I'll] roast it. ' maybe roast-NONPASf

As can be seen in the above examples, nguba always precedes the expressed assumption or alternative.

4.10 Interjections

Interjections usually make up a complete utterance, or else precede a sentence. In the latter case they are set off from the rest of the utterance by a slight pause. Some interjections are monosyllabic and/or end with a glottal stop or breathy voice. In this respect they differ from the normal phonology and word pattern in Kuku Yalanji. Interjections with an approximation of their meaning are:

yuwu	'yes'
kari	'no' (both used in reply to a question)
ba	'come on, hurry up'
ma	'go on, go ahead, don't let me keep you'
kaku	'just a minute, wait for me'
ngay	'oh yes, I see' (used as an affirmative noise by the interlocutor in conversation)
yalada(?)	'that's all right, that's settled then' (typically concludes a conversation, often ending with a glottal stop)
yala	'OK, let's' (in this sense it typically precedes a suggestion; but it may also be used like yalada)
juyuy	'that's nonsense, take no notice of it'
wumba	'never mind'
yah	'hey' (used to attract attention)

5 Narrative discourse

5.1 Preliminary remarks

Above the sentence level any language displays regularities in the way in which topics are introduced, reference is maintained, information is ordered, turn-taking is achieved, feedback is prompted and given and many other aspects of discourse are handled. Brown and Yule (1983:22) define a "regularity" in discourse as "a linguistic feature which occurs in a definable environment with a significant frequency".

This chapter is concerned with linguistic features that fulfil certain discourse functions in Kuku Yalanji narratives. The narrative has been chosen as an unambiguously definable environment, i.e. a text in which the topic is identified by the narrator ("this is a story about ... "), and there is a beginning, purposeful progression through a plot, and an end. Within such a framework regularities can be identified and compared across several texts. ! Neither conversational structures nor any other genres of discourse are investigated here.

In these narratives I have concentrated on cohesion, choice and marking of topic, and the distinction between events and non-events. This choice was guided partly by concepts provided in Grimes The thread of discourse (1 975), but more importantly by the need to explain the function of varying word order, aspects of noun, pronoun and zero reference and the use of some particles and clitics.

5.2 Cohesion

From the listener's point of view it is of foremost importance that a narrative is cohesive, so that there is no difficulty in understanding which participant does what, where and when. This necessary cohesion includes several aspects: participant identification and progression of the narrative through time and space, as well as the rate at which it progresses.

Halliday and Hasan (1 976) regard cohesion as a semantic concept in the same way as text is a semantic unit. Cohesion is represented in the interdependence of the constituents of a text: "Cohesion occurs where the INTERPRET A TION of some element in the discourse is dependent on that of another" (Halliday & Hasan 1 976:4).

See Brown and Yule ( 1 983 :68ff.) on identifying analysable chunks of discourse.

202 Chapter 5

5.2. 1 Participant Identification

Participant identification first of all has to establish a reference sufficiently well that the audience knows who or what is being talked about, and then has to maintain reference in a manner that prevents confusion on the part of the listener.

Generally speaking, reference maintenance is the domain of pronouns, while nouns are used predominantly to establish reference but may also appear in anaphoric function. Kuku Yalanji frequently combines nouns and pronouns with the pronoun either preceding or following the noun. In keeping with the general convention mentioned above, a pronoun-noun sequence can be assumed to have anaphoric or definite reference, e.g. nyulu jalbu 'the woman' in contrast to a noun on its own, e.g. jalbu '(a) woman' which usually conveys new andlor indefinite reference. A noun-pronoun sequence, on the other hand, normally establishes a new reference which is immediately followed up by an anaphoric element, e.g. jalbu nyulu 'woman, she'. The following sub-sections deal with the application and modification of these basic possibilities.

5.2. 1. 1 Establishing participant reference

The need for detailed identification varies in Kuku Yalanji, and presumably in other languages as well, according to whether the referent is first, second, or third person, singular or non-singular. Reference to first and second person singular and first person dual inclusive participants is established and maintained in Kuku Yalanji discourse simply by the use of pronouns ngayu 'I', yundu 'you sg. ' and ngali 'you and I', since obviously there can be no doubt about the identity of the referent(s).

However, when introducing any participant other than those mentioned above, the speaker will provide additional information in the form of one or more proper or common nouns which may be preceded or followed by an appropriate pronoun. For instance:

Third person singular:

(611)	Pastor nyulu ...	'Pastor, he ...' (text 3, line 9)
(612)	John Robertamu kangkal ...	'John Robert's child ...' (text 3, line 34)
(613)	Wulbuman ...	'An old woman ...' (text 51, line 15)

First person dual exclusive:

(6 1 4) Ngaliny yaba-ngka kuyu wubul 3du.exc.NOM(A) eB-ERG:pt(A) fish.ABS(O) many.ABS(O) 'My elder brother and I caught a lot of fish.' mani-ny. get-PAST

Second person dual:

(6 1 5) Yubal 2du.NOM(S) Roderick janji! Roderick.ABS(S) swim. IMP 'You and Roderick have a swim!'

Third person dual:

(6 1 6) Warru-bulal jambul, bula ... yg.man-pair.ABS(S) two.ABS(S) 3du.NOM(S) 'A pair of two young fellows ... ' (text 51, line 1) First person plural inclusive:

(6 1 7) Ngana kangkal-kangkal jilba-dunga-y wunkuny. Ipl.inC.NOM(S) child-REDUP.ABS(S) walk-go-NONPAST tomorrow 'You and I and the children will go walkabout tomorrow. '

First person plural exclusive:

• (6 1 8) Ngayu babi I sg.NOM(S) FM.ABS(S) wulbuman yindu ngamu old.woman.ABS(S) other.ABS(S) mother.ABS(S) nganjin dunga-ri-ny mayi baka-nka. Ipl.exc.NOM(S) go-PLS-PAST food.ABS(O) dig-PURP 'I, grandmother, another old woman and mother, we went out to dig for food (yams). '
• (6 1 9) Nganjin Mabel balka-n-balka-wa-ny. ¹pl.exc.NOM(S) Mabel.ABS(S) speak -n-REDUP-RECIP-PAST 'We all including Mabel had a long chat. '

Second person plural:

(620) Yurra karrkay dunga-y bana mana! 2pl.NOM(S) child.ABS(S) go-IMP water.ABS(O) get.IMP 'You children go and get water!'

Third person plural:

(621)	Jalbu-jalbu	karrkay-karrkay	jana	wawubaja-nga
	woman-REDUP.ABS(S)	child-REDUP.ABS(S)	3pl.NOM(S)	river-LOC

dunga-ny mukirr mani-nka. go-PAST mussel.ABS(S) get-pURP 'Women and children, they went to the river to get mussels.'

(622) Jana Doris kada-ny-baja. 3pl.NOM(S) Doris.ABS(S) come-PAST -again 'Doris and her people came back.'

Constructions of this type are typically used when introducing participants for the first time in a discourse. The frequent combinations of nouns and pronouns can be explained as follows: In examples (6 1 1), (6 1 6), and (62 1 ), all referring to third person participants, the pronoun follows after identification has been established through nouns and the pronoun has anaphoric function. (Incidentally, the tautology in (6 1 6), "a pair (of) two", is a common device when introducing a couple of participants who are thought to belong together.) On the other hand, in examples (6 1 4), (6 1 5), (6 1 7), (6 1 9) and (620), all referring to first or second person participants, the pronoun appears initially as if referring to someone already known. This is indeed the case, since at least one referent in these constructions, i.e. the speaker or the addressee, is defined by the very pronoun. Only where the hitherto unidentified members of a group far outnumber the known member, and are considered relevant by the speaker, is the group pronoun preferred to follow the list of participants as in (6 1 8). Conversely, where one member is deemed sufficient to identify a whole group in a third person plural reference, the pronoun is preferred in initial position, followed by the identifying noun, as in (622) which conveys something like 'Doris and her usual hangers-on, you know who I mean'.

The above examples are the most common conventions for introducing new participants in discourse. Apart from these, a speaker is free to choose other forms for stylistic reasons. He or she may for instance introduce the first character in a narrative, usually if it is the most important one, straight away by what is normally anaphoric reference, as for instance:

• (623) Nyulu karrkay wunanany. 'The child was sleeping.' (text 36, line 1)
• (624) Nyungu bayan wangkar wuburrbu. 'His camp was on top of the mountain.' (text 1 2, line 3)

By this means the speaker quickly draws the audience into the narrative by implying immediate familiarity between the listener and the main character.

Another device is the use of cataphoric demonstrative yanya 'this', as for example in text 51, line 72:

(625) Nyulu wulman yanyabi bundandany. 'There was this old man.'

The old man is introduced as a new participant in this form after about two-thirds of the narrative and because of the cataphoric demonstrative we can expect him to do something important, which he does. He reassembles the dog skeleton and resurrects the animal; the old man is in fact the hero of the story.

5.2. 1.2 Maintaining participant reference

Kuku Yalanji has a variety of means by which to maintain reference throughout a narrative, including nouns, pronouns, demonstratives, or no overt reference at all. In the previous section we have seen that demonstrative yanya 'this' is used to introduce a new participant with cataphoric implication. Demonstrative yinya 'that', on the other hand, refers back to an already known participant. It is generally used to reintroduce with special emphasis an important participant after the narrative had focused on someone or something else for a while. For instance in text 36 the first text line introduces the main character, the child, with definite reference: Nyulu karrkay wununany 'The child was sleeping'. The next two lines mention that mother and father were also asleep, implying that it was fairly late at night and all was peaceful and quiet. Line 4 then reverts to the main character:

(626) Yinya karrkay nyulu bijarriny. 'That child, he had a dream.'

Later on, the narrator describes how the people went to the river and dived for mussels and then again re-focuses the narrative onto that particular child who had had the dream (lines 25-26):

(627) ... yinya karrkay bada. Yinya janjiny. ' ... that child (went) down. That one dived.'

Thus we may say that demonstratives in Kuku Yalanji have dual function in simultaneously referring to both a participant and his/her prominent role.2 Yanya 'this' establishes this twofold reference, while yinya 'that' maintains it.

² See also use of demonstrative 'that' in Guugu Yimidhirr discourse as outlined by Haviland (1979a: 1 56-1 57).

Nouns, pronouns and zero anaphora maintain a less marked reference. It would seem that the use of pronouns alone is possible only where these are sufficient to distinguish between participants - for instance if first person singular ngayu interacts with third person singular nyulu, or third person singular interacts with third person dual bula - but not where two or three third person singular participants are involved. In the latter case nouns can be used for unambiguous reference if switching from one participant to another, e.g. nyulu warru 'the young man' or nyulu wulman 'the old man'. However, Kuku Yalanji narrators can get by with very little nominal reference, indeed with very little overt reference at all, and still be understood.

The main key to referent identity, after it has been established, lies in the context given by the narrative and specifically in contextually appropriate actions performed by the participants. One basic rule is that as long as a common participant is maintained, be it in S, A or ° or even oblique function, no overt reference is required. Thus, if we encounter a stretch of text in Kuku Yalanji narrative where participant reference is omitted, we can safely assume that the participant which had been stated at the beginning of this stretch is involved throughout. For instance text 36, lines 1 1-13; line 11 introduces the participant:

(628) lana bama jabajaba wandiriny .. . 'A whole lot of people turned up ... '

followed by zero reference in the next line. But it is clearly the people, and not the child mentioned earlier, who take torches, flash them about and look under the bed. Another example is in text 3, lines 34-37. Lines 34 and 35 identify the intruder. Line 36 describes an action: Yirrkany 'Called out.' It would be possible that the people in the camp called out to the stranger, but from the zero reference we can infer that it is in fact the stranger who does so. Line 37, again a new clause with zero reference, explains how the stranger got there: Motorbikemunji 'With a motorbike. '

As mentioned above, a contextually appropriate action can be a means of identifying an otherwise unspecified participant. This aspect closely interacts with the principle of zero reference = same participant. Take for instance text 51, lines 49-5 1. The dog (kaya in line 49) runs on and the two fellows (bula) spear it in line 50. There is no reference to the dog in ° function, but it is the dog, and not something else, that got speared. In line 51 the dog is cut up into halves and there is no reference to either A or 0, but there can be no doubt as to who is in what role.

The convention of contextually appropriate participant can even identify a participant in a switch-reference situation with zero anaphora, in which case it overrides the other convention of zero anaphora = same participant. Consider the first part of text 36 where, after initial identification, the participants, the child and the people, are distinguished largely by their actions: the child dreams and screams and the people look for the imaginary crocodile. In lines 1 5-1 8 the child's anguished cries are quoted, followed by a narrator's comment on the intensity of the child's dream in line 19 (zero anaphora = same referent). In the following line there is still no overt reference to an agent, but the verb balbanybaja 'flashed a light again' signals a switch in reference from the child to the people who, as we know, have previously been using a torch. Thus the notion of 'appropriate participant for a role' can override the convention of 'zero anaphora = same participant'.

It appears that these conventions of reference maintenance through contextual appropriateness are the backbone of participant cohesion in a Kuku Yalanji narrative.

5.2. 1.3 Summary

To summarise the most essential points, participant reference in a first introduction is normally established by indefinite nominal NPs, but a definite construction, including cataphoric demonstrative, may be used to imply a certain importance of the referent.

Maintenance of participant reference most commonly involves pronouns or zero anaphora. Where pronouns are indistinct in third person reference and where zero anaphora is applied, participants can be identified according to the following conventions:

(i) zero anaphora/indistinct pronoun = same referent

(ii) zero anaphora/indistinct pronoun = contextually appropriate participant.

Of these, convention (ii) overrides convention (i).

5.2.2 Temporal cohesion

Narratives are typically recounted in the past tense in Kuku Yalanji and normally the order in which events are told represents the chronological order in which they take place. However, the sequence of events in a narrative cannot always be strictly linear, since, as in real life, actions by different participants may overlap.

Simultaneous subordinate clauses (see §4.4.4.1) can be used to indicate that actions by two participants occur at the same time. However, the application of simultaneous subordinate clauses is limited to only a very brief time span. If actions by different participants overlap through any length of time, the narrator switches back and forth between them in order to bring the diverging actions 'up to date'. Such a sequence can be found in text 51, lines 37-50. In line 37 the old woman goes off to fetch her dog, as indicated by a purposive clause. From this we can infer that she is occupied for a while as the narrator turns our attention back to the two fellows who are said to take off and run westwards. While the two young men are left running, the narrator switches back to the old woman, who fetches the dog and sets it on the trail of the two young men (lines 40-41 ). The dog then takes over in an active participant role (the old woman drops out of sight) and keeps following the young men. From then on, line 46, participant reference switches with every sentence in "leapfrog fashion" (see Dixon 1 972:7 1) until the unity of time and place is re-established in line 50 where the two fellows kill the dog.

In addition to these systematic reference switches, the clitic -kuda (TEMP2) 'meanwhile' (as in lines 38 and 39 in the above sequence) apparently plays a role in indicating overlapping actions. The function of -kuda is still not entirely transparent3 but its interpretation as 'meanwhile' appears appropriate in many examples, for instance in text 51, lines 38 and 39 and text 3, lines 16 and 1 8, or the following lines from a text recorded by Hershberger and reproduced in a booklet Kuku Ngujuji. (This text does not contain glosses and translation is in the form of an English adaptation; glosses and translation are therefore my own.)

(629) Babi-ngka juwal-da-ku-Iu wadu-li dama-ny. FM-ERG:pt(A) digging.stick-INST-EMPHI-EMPH2 wrong-EMPH2 spear-PAST 'Grandmother missed [the fish] with the digging stick.'

³ H. Hershberger and R. Hershberger (1982:1 93) analysed -kuda as 'definitely, for sure', giving the example Nyulu dungaykuda 'He'll go for sure'. However, no context is given for this utterance.

(630) Ngayu-ji-ku-lu naybu-bu-ku I sg.NOM(A)-in.tum-EMPHI -EMPH2 knife-INST-EMPHI

bana-nga-kuda wadu-kuda nyanda-ny. water-LOC-meanwhile wrong-meanwhile chop-PAST 'While I in tum hacked vainly in the water with a knife.'

Another way to indicate temporal relations in a narrative is the use of temporal clitics -da 'then', -(ng)Vrr 'first of all' and wawu- 'after' (see §3.10.1). Clitic -da indicates that an action follows one or more others, for example:

(63 1) Nganjin wawubaja-nga kada-ny tent ngara-ny IpLexc.NOM(S) river-LOC come-PAST tent.ABS(O) set.up-PAST mayi nuka-ny wuna-ny-da. food.ABS(O) eat-PAST lie-PAST-TEMPI

'We came to the river, set up tent, ate some food and then rested.'

In text 36, for instance, lines 37-38 describe the emergence of the crocodile which grabs the child (line 39) and then takes it underwater:

(632) Karrba-ny, wundi-ny-da,janji-ny-da '[It] grabbed [him], then took [him with it] and dived.' (text 36, line 40)

Clitic -da may be used to close a sequence of events after which either a new participant acts (e.g. the child's parents in text 36, line 41), or some time may elapse before the plot continues. Example (63 1) above, for instance, could be followed by 'Next morning they got up, etc.'.

Clitic -(ng)Vrr, on the other hand, stresses the first action or participant in a sequence, for example:

• (633) Ngayu kuyu waju-l-urr wundi-l-da. I sg.NOM(A) fish.ABS(O) cook-NONPAST-TEMP3 take-NONPAST-TEMPI 'I first cook the fish and then I take it [with me].'
• (634) Queenie-ngka nyulu-ngurr diburr Queenie-ERG:pt(A) 3sg.NOM(A}-TEMP3 egg.ABS(O) 'Queenie was the first to find (lit. get) an egg. ' mani-ny. get-PAST

Clitic wawu- in conjunction with a nominal in ablative case indicates 'after, since', for example:

• (635) Wawu-warngku-muny jana TEMP4-sleep-ABL 3pLNOM(S) 'After the sleep they went on.' dunga-ny-baja. gO-PAsT-again
• (636) Wawu-yinya-muny ngayu yalbay-ma-ny. TEMP4-that-ABL I sg.NOM(S) big-INCHO-PAST 'Since that time I have grown up. '

Time spans during which nothing in particular happens are always bridged by specific temporal reference, such as wunkuny 'next day', muduwajuny 'at daybreak', etc. (for time words see §3.4).

208 Chapter 5

5.2.3 Anaphoric linkage

Anaphoric linkage, as defined by Grimes (1975:3 16), is the repetition or paraphrasing of a clause "in dependent form as a means of leading into an independent clause that contains new information". Kuku Yalanji narratives frequently show repetition of clauses or parts of clauses, although not in dependent but rather co-ordinated structures. Clauses may even be repeated in non-co-ordinated form in which case they often show a change in word order. Consider text 51, lines 4-7:

(637) 4	Bula yuwulbuwun kada-ny. 3du.NOM(S) Jeannie.River come-PAST 'They came to the Jeannie River.'
5	Bula kada-ny, bula wawubaja-nga dunga-ny. 3du.NOM(S) come-PAST 3du.NOM(S) river-LOC go-PAST 'They came and they went to the river.'
6	Bula wawubaja-nga dunga-ny, bula mayi 3du.NOM(S) river-LOC go-PAST 3du.NOM(A) fruit.ABS(O) kabal warrmba-bunga-ny. Leichardt.tree.ABS(O) uncovered-CAUS-PAST 'They went to the river and they found some Leichardt tree fruit.'
7	Kabal mayi warrmba-bunga-ny. Leichardt.tree.ABS(O) fruit.ABS(O) uncovered-CAUS-PAST 'Leichardt fruit [they] found.'

Even though such sections of text in Kuku Yalanji do not involve dependent structures, the typical repetition of information presumably justifies the use of the term 'anaphoric linkage'. Further examples of anaphoric linkage may be found in text 51, lines 89-9 1 and 97-98, and text 36, lines 39-40.

Anaphoric linkage in Kuku Yalanji narratives is not strictly correlated with the distinction between events and non-events.4 But while it may occur anywhere in a text, its use is apparently favoured in certain situations. Firstly, at the beginning of a narrative (as in the quoted example (637) above), perhaps with the intention of slowly 'warming up' the audience rather than overwhelming them with too much new information. Secondly, it is used immediately preceding a specially important event where it seems that the narrator wants to stall the flow of information to build up tension; and thirdly, at the peak of an event to provide the crucial happening with more impact, as for instance in text 36, when the crocodile grabs the child, or in text 51, when the dog comes back to life.

5.3 Topics

5.3.1 Preliminary remarks

Grimes ( 1 975:323ff.) has attempted to clarify the somewhat confused terminology of 'topic', 'focus', 'theme' and 'emphasis' in linguistic literature. To this end he has introduced

⁴ Grimes ( 1975:96) mentions that in Kayapo of Brazil anaphoric linkage is used exclusively for nonevents, so that the auclience is alerted to an important section in the narrative as soon as anaphoric linkage ceases.

the theatrical term 'staging' to refer to the process of presenting a narrative, or parts thereof, from a particular perspective (see Brown & Yule 1 983: 1 34). The speaker arranges his narrative like a stage performance with some participants acting at the centre of the stage and others off-centre, or even leaving the stage and perhaps re-appearing. The decision of who is to be at the centre of the stage is the semantic choice of theme and the practical implementation of this choice is "the designation of a constituent in the grammar as the TOPIC by means of appropriate signaling devices" (Grimes 1 975:324). The continuing and overall process of staging involves the choice of a theme whose overt surface form is the grammatically marked topic in a stretch of discourse. Thus 'theme' is a discourse or semantic term whereas 'topic' is a grammatical one. Since grammatically marked topics constitute the 'theme' in a narrative the grammatical devices for topic marking will be investigated in the following subsections.

5.3.2 Topic choice and topic marking

In the following subsections I will look at how topics are represented, and at which is the most common choice of topic in Kuku Yalanji. An investigation of these two questions could easily become circular: can one define a topic by the way in which it is marked or does one find out about topic marking by assuming that certain constituents are topic? The topic as part of a discourse theme is the most prominent or salient feature in a sentence. "Clearly the marking of thematization is related to a semantic factor of PROMINENCE" (Grimes 1 975:327). But how do we know which feature is the most prominent unless the speaker conveys this in some way? The best starting point for an investigation of this problem appears to be the universally applicable feature of prominence in word order.

5.3.2. 1 Word order and syntactic function

"Many of the phenomena written off in grammars as free word order are in contrast with each other thematically" (Grimes 1 975:331). Kuku Yalanji, like most Australian languages, has 'free word order', even to the extent that noun phrases may be split up and their constituents distributed throughout the clause. Nevertheless there is a preferred or unmarked word order, which is S/A 0 V. But variations of this unmarked word order are not uncommon and we must assume that this has thematic implications.

According to a universal convention, the leftmost position in a sentence is the most prominent one.5 This means that leftmost position is the natural place for the topic as the most salient constituent. Considering that the preferred word order in Kuku Yalanji is S/A 0 v, the normal choice of topic would therefore be a noun phrase in S or A function. (These functions coincide in form for pronouns but are distinct for nouns.)

Since Kuku Yalanji allows variation in word order, it stands to reason that an 0 NP can be topicalised by simply moving it to leftmost position. This is indeed so, but apparently there must be a good reason why an 0 NP, or even an oblique noun phrase, is to be topicalised; i.e. speakers are not indiscriminate in their choice of what is to be topic. Consider the following examples:

210 Chapter 5

(63 (58) Nyulu jina yijarri-ny 3sg.NOM(A) fOOt.ABS(O) put-PAST murraJa-nga. stonefish-LOC

Nyunguniny murraja-bu baka-ny. 3sg.ACCCO) stonefish-ERG(A) poke-PAST 'He put [his] foot on a stonefish. The stonefish poked him.' (From a tale about how lina-baji 'sore foot' acquired his nickname.)

(639) text 3, line 44

Chris	kangkal-bulal	dingkar-bulal	ngayku
Chris(O)	own.child-pair.ABS(O)	man-pair.ABS(O)	1sg.POSS.ABS(O)
kangkal-kangkal,	ngayu	janangan	murrukanga-ny.
own.child-REDUP.ABS(O)	1sg.NOM(A)	3pl.ACC(O)	scold-PAST
'I scolded them, Chris and my two sons:'

(In the latter example, reprise (see §5.4.1) is used to front the 0 NP.)

The above examples are taken out of contexts in which the referents that are in 0 function here had already been established as topic, in S or A function, in the preceding sentences. Furthermore, in (638) there is the additional aspect of the 0 NP referring to a human being, whereas the agent is lower in animacy. (This point will be discussed in the next subsection.)

Such occurrences of 0 NP in leftmost position as topic must not be confused with instances where zero anaphora applies to the agent. If the 0 NP is in leftmost position in such situations, the unmarked word order is in effect maintained, but the 0 NP is in initial position by default because the agent has zero reference.

We can thus determine two conditions under which the choice of 0 NP as topic is most likely:

• (i) If the referent of the 0 NP is of higher animacy than the referent of the A NP;
• (ii) If the referent of the 0 NP has previously been established as topic in another function.

Nevertheless, there seems to be a general reluctance to use an 0 NP as topic. To avoid this, Kuku Yalanji speakers may resort to a passive construction which places the patient in S function, as for instance in text 12, line 53:

(640) Kanbal kuni-ji-ny... some.ABS(S) kill-ITR-PAST 'Some were killed ... '

Another way to avoid having an 0 NP as overtly marked topic is zero anaphora in a switch function situation (see below).

5.3.2.2 Switch function

In §5.2.1 .2 we have seen that participants can be referred to by zero anaphora throughout longish parts of a narrative. The criteria for participant identification in such cases are 'same referent' and 'contextually appropriate participant'. This section discusses the syntactic conditions under which the same participant can be referred to by zero anaphora.

As we have already seen in §5.3.2.1, it is uncommon for an 0 NP to appear overtly in topic position in an anaphoric situation, and there is not a single occurrence of a topic being

introduced as transitive object in the appended texts. We can assume that a new topic is usually introduced in S or A function, which coincides with the normal choice of topic in a clause with unmarked word order. Once a topic has been introduced, zero ana ph ora can apply, even if the referent appears in different syntactic functions. However, as is evident from the texts, a topic that is introduced in s function is far more likely to be followed by a sequence of zero anaphora in the same or different functions, than is a topic that is introduced as A.

I have looked at 50 topic introductions followed by one or more instances of zero anaphora. Of the 50 introductions 35 are in S function and 15 in A function. As can be seen in Table 5.1 there is evidently no problem for a topic that has been introduced as S to be followed by several zero references in all syntactic functions (20 occurrences of S, 20 occurrences of A, and 2 occurrences of 0). Again, a topic in 0 function is quite rare.

Function of topic in introduction		Function of topic in subsequent clauses with zero anaphora
		S	A	O
S	35	20	20	2
A	15	3	12	-

Table 5.1: Syntactic functions of topic in introduction and subsequent clauses with zero anaphora

If, on the other hand, a topic is introduced as A, subsequent zero anaphora is apparently restricted. It can apply freely where the referent remains in the same syntactic function (12 occurrences), but is rare where the referent assumes different syntactic function (3 occurrences of S, no occurrence of 0).

Like switch reference (see §5.2.1 .2), switch function does not require any particular grammatical devices in Kuku Yalanji. Again, Kuku Yalanji differs in this respect from some other Australian languages. In Dyirbal, for instance, a switch from S or 0 to A must be indicated by antipassive with verbal derivation -lJay-, and a switch from A to S or 0 involves a special verbal inflection -lJura (Dixon 1 972:7 1-74 and 77-79 respectively). As Silverstein (1976: 1 56) remarks: "The markers -lJay- and lJura, 'then', are discourse markers that show the switch of underlying function of co-referential noun phrases".

To summarise, on the evidence of topic introduction with subsequent zero anaphora for same or different function, we may say that the first choice of topic in Kuku Yalanji is a noun phrase in S function and second choice is a noun phrase in A function. This agrees with Haviland's (1 979a:1 59) statement for Guugu Yimidhirr: "s NPs are unambiguously available as topics and can be linked with S, A or 0 NPs in second and subsequent clauses".

5.3.2.3 Choice of referent as topic

The choice of a particular referent as topic correlates on the one hand with the preferred syntactic function of the topic as S or A, and on the other hand with the notion of animacy versus mammacy.

Unlike nouns, pronouns, which refer exclusively to animate beings in Kuku Yalanji, and particularly to humans, are not formally distinct for S or A function. That is to say that pronoun forms are distinguished as topic versus non-topic and are independent from the verb, whereas noun forms are determined by the verb according to whether it is transitive or intransitive. This would suggest that those referents that are always denoted by a pronoun, i.e. first and second person pronoun referents, are most likely to be topic if they occur in a narrative.

Further, we have seen that in the less frequent cases where a lower animate or even inanimate referent as agent acts upon a higher animate referent as transitive object, the 0 NP is preferred in leftmost, i.e. topic position, as in (638) above, or the clause may be passivised to transform the transitive object into S and therefore natural topic, as in (640). Such formations are also the preferred form in elicited sentences with a lower animate agent acting upon a higher animate patient.

We can thus set up the following hierarchy of choice of referent as topic:

• (i) first and second person pronoun referents
• (ii) animate referents
• (iii) inanimate referents.

5.3.2. 4 Conclusions

Following the above findings, some speculations can be made on the classification of Kuku Yalanji within the framework of topic prominent (Tp) and subject prominent (Sp) languages developed by Li and Thompson (1 976). This framework includes a diachronic schema for topic prominent and subject prominent languages with the following stages (Li & Thompson 1 976:485):

A Topic-prominent language

Topic notion integrated into basic sentence structure; topic and subject are distinct.

B Neither subject-prominent nor topic-prominent

Topic becomes more closely integrated into case frame of verb.

C Subject-prominent

Topic has become integrated into case frame of verb as subject; subject and topic are often indistinct, subjects have some non-topic properties; sentences with clear topics are highly marked.

D Both topic-prominent and subject-prominent

Topic sentences become less marked and more basic. (The next stage of development after this is again A.)

Clearly, there is a correlation between topic and grammatical subject in Kuku Yalanji which means that it is not a topic prominent language. However, the grammatical subject in Kuku Yalanji has two distinct and clearly marked functions for nominal NPs: intransitive subject (S) and transitive subject (A). Of these S is favoured as topic over A, although A is a possible choice under more restrictive circumstances. Furthermore, sentences with zero anaphora for the topic do not have to be clearly marked in Kuku Yalanji; they can be

transitive, even though the topic is introduced in S function. Compare this with Dyirbal, a subject-prominent language according to Li and Thompson ( 1 976:460), where a clause with topic in A function following a clause with the same topic in S function must be intransitivised through antipassive so that the second function of the topic is identical with the first (Dixon 1 972:7 1-74).

Therefore, because topic correlates with grammatical subject in Kuku Yalanji, with a preference for intransitive subject over transitive subject as topic, and because topic sentences do not have to be specifically marked, I suggest that Kuku Yalanji is both topic-prominent and subject-prominent, according to the Li and Thompson classification.

5.4 Localised highlighting

A Kuku Yalanji speaker may emphasise any single clause constituent that he or she wishes to bring to the listener's special attention, either by a change in word order or with the help of emphatic clitics.

5. 4. 1 Specifically marked word order

Any clause constituent may be placed in prominent leftmost position for special emphasis. An example of this is in text 1 2, lines 40-42:

(641) 40	Nyulu	kulji	yalbay-ku	mani-ny.
	3sg.NOM(A)	rock.ABS(O)	big.ABS(O)-EMPH1	get-PAST
	'He got a really big rock.'

• 41 Kulji yalbay-ku jurrnga-ny, yijarri-ny. rock.ABS(O) big.ABS(O}-EMPHl push-PAST put-PAST '[He] pushed the really big rock and set it up. '
• 42 Baral-ba-ku yijarri-ny kulji yalbay. road-LOC-EMPHI put-PAST rock.ABS(O) big.ABS(O) 'Right on the road [he] set the big rock.'

(This example also displays the emphatic clitic -ku, to be discussed in the next sub-section.)

Line 40 has unmarked word order. Line 41, with zero anaphora for the topic, leaves kulji yalbayku in leftmost position, which helps to alert the audience to the significance of the rock. Line 42 begins with baralbaku 'right on the road' to emphasise the implications of this trap that is so ingeniously constructed by the Aborigine.

Verbs, adverbs and adjectives may also be fronted for emphasis. An example of a verb in clause initial position is in text 1 2, line 25:

(642) Bundanda-ny nyulu nyaji-l-nyaji-ny. sit.REDUP-PAST 3sg.NOM(A) see-l-REDUP-PAST 'He sat and watched.' ("Sitting down, he watched.")

This stresses the composure of the Aborigine in the face of the approaching group of police.

Apart from simple fronting for emphasis, reprise constructions (see Grimes 1 975:342) are frequently used in Kuku Yalanji discourse. Reprise involves the statement of a noun phrase

in initial position, before the clause as it were, with an anaphoric pronoun following in the slot where this NP would normally occur, for example:

Reprise of S NP:

(643)	Waybala	janyiman,	jana	Palmer River	kada-ny
	white.man.ABS(S)	Chinese.ABS(S)	3pl.NOM(S)	Palmer.River	come-PAST
	gold
	gold.ABS(O)	mani-nka.
		get-PURP
	'White men and Chinese, they came to the Palmer River to get gold.'

Reprise of A NP:

(644) Waybala-ngka, jana bama white.man-ERG:pt(A) 3pl.NOM(A) Aborigine.ABS(O) kuni-l-kuni-ny. kill-l-REDUP-PAST 'White men, they killed many Aborigines. ' wubul many.ABS(O)

Reprise of 0 NP:

(645)	Bama Aborigine.ABS(O)	wubul, many.ABS(O)	waybala-ngka white.man-ERG:pt(A)	janangan 3pl.ACC(O)
kuni-l-kuni-ny.
kill-l-REDUP-PAST
'Many Aborigines, the white men killed them.'

Another example of reprise applying to an 0 NP was given in sentence (639).

5 .4.2 Emphatic clitics

The semantic distinctions of emphatic clitics -ku (EM PH 1), -lu (Yalanji), -la (Nyungkul) (both EMPH2) and -bi (EMPH3) are not entirely clear. They appear to have different emphatic strength and play different roles in the interaction between speaker and listener.

The form -ku is the most frequently occurring emphatic clitic and therefore appears to be a rather commonplace emphatic marker (similar to 'really' in English). The assumption that it is relatively weak is supported by the fact that it may be followed by other emphatic clitics lu and -bi (see for instance -ku-lu in examples (629) and (630).

Clitics -La in the Nyungkul dialect and -lu in the Yalanji dialect are similar in form and have partly similar functions. Yalanji -lu is used predominantly to mark whole parts of text as particularly important events, which will be discussed in §5.5.1.1 . Nyungkul -La does not have this function, but both dialects may use -la or -lu to emphasise individual nouns. These clitics seem to request the audience to pay special attention and could be translated as 'hark!' or 'pay attention!'

A typical use of -lu in this function occurred in conversation and is worth while recounting here: Mrs Mabel Webb had begun to tell me about a trip that she had undertaken together with some other women, when she sneezed. I asked whether she had a cold: "Yundu ngikiji?" Mrs Webb misunderstood me as saying: "mijiji? ", i.e. enquiring whether the people involved in her tale were white women, and she replied somewhat impatiently: "Kari,

bamalu!" "No, they were Aborigines (do pay attention to what I say)!" Nyungkul -La is exemplified in text 51, line 97, repeated below:

(646) Nyulu warndiny kayala. 'It got up, the dog [did].' (fhe dead dog well and truly got up again - note this, listener!)

While -ku and -lui -La could be seen to assert the significance of something relative to the context of the plot, clitic -bi may have its function more as an affective marker in the interactive context in which a story is told. H. Hershberger and R. Hershberger (1 982:294) suggest -bi to indicate "surprise about something in the narrative". The listener is invited to share a sense of amazement about some event. Consider some of its uses in the appended texts:

(647) text 36, line 37 yinyabi bijarrmuny

'that one from the dream!'

(648) text 36, line 53

yala bijarrbi 'just like the dream!'

In both these examples -bi emphasises the astonishing fact that the events in real life tally with the child's dream.

(649) text 51, line 72

Nyulu wulman yanyabi bundandany ... 'There was this old man ... ' (who is he? what's he up to?)

In this example the speaker's signal of surprise could be regarded as aptly adding emphasis to the sudden coming into focus of this new (and very important) participant.

(650) text 3, line 6 bubu julbarrbi 'the ground was [extremely] slippery!'

Here, we may infer from clitic -bi that the road was a good deal more slippery than the narrator's party had expected.

5.5 Events and non-events

"The first distinction made in the analysis of discourse is between events and non-events" (Grimes 1 975:35). Events represent what happens in a narrative, whereas non-events provide information about where, when, how or why something happens. The following sections concentrate on grammatical devices used in Yalanji and Nyungkul to set off events from non-events; the use of lexical verbs to indicate spatial setting; and the function of some particles in contrasting what did not happen from what did happen.

5.5.1 Basic distinction between events and non-events

Nyungkul speakers and Yalanji speakers have different ways of distinguishing between events and non-events. Nyungkul speakers indicate non-events by using a specific verb inflection. Yalanji speakers, on the other hand, mark particularly important events with an emphatic clitic.

5.5. 1. 1 Important events in Yalanji dialect

In the Yalanji dialect a narrator may make copious use of the emphatic clitic -lu to mark events deemed to be particularly important. This clitic has been defined above (§5.4.2) as an attention-getting device. It may attach to all kinds of clause components and often follows the general emphatic clitic -ku.

Let us look at the excerpts from text 3 which contain the first two -lu-marked events in a very long and meandering narrative. Previous to line 1 of the excerpt Mrs Walker had related how her party had prepared for the trip. Lines 1-14 tell us how the group is driving along in the rain. In lines 1 5-1 7 there is the first significant event: they encounter difficulties on the slippery road, the driver tries to control the car, but the car keeps slipping about in a most alarming fashion. Clitic -lu occurs three times, always with verbs and in conjunction with -ku. Once this problem is overcome, the journey proceeds 'uneventfully' and no -lu is used until the event involving the strange visitor on the campsite. But it is not really the appearance of the stranger that is marked as significant, rather the unsatisfactory behaviour of the young male members of Mrs Walker's party. In lines 43-49 -lu is used six times, always in connection with nouns. In the quotations in lines 45 and 49 -lu is used probably more in the function of localised highlight, nevertheless it occurs within the boundaries of this particular event. (The most frequent occurrence of -lu that I have encountered is 12 times in 10 text lines, preceded by 28 text lines in which -lu appeared only 5 times and widely apart as local emphatic marker.)

To sum up, we can say that emphatic clitic -lu in Yalanji, when used repeatedly throughout a section of text, marks this section as a significant event in contrast to others. Sequences marked with clitic -lu may be regarded as the 'high points' or 'peaks' in a Yalanji narrative (see Longacre & Levinsohn 1 977: 1 09).

5.5. 1.2 Non-events versus events in Nyungkul dialect

While Yalanji uses a special device for marking peaks, Nyungkul applies a certain morphological feature, the unmarked verb inflection -nyV, for non-events. At sentence level, suffix -nyV functions as subordination marker (see §4.4.4). However, it also occurs frequently in texts, in strings of verbs or several consecutive sentences, where there is no evidence at all for subordination; i.e. its function cannot be explained in syntactic terms. In such an environment -nyV appears to indicate 'non-event', a discourse function.6

Consider text 51 which contains in lines 61-67 an explanatory remark on how people prepared ground ovens in the old days. This part of text was clearly set off as a parenthetic remark by the speaker's intonation and voice quality. In other words, he interrupted the story

⁶See Grimes' ( 1975:93) observation that the aspectual system of some languages can be fully understood only within the framework of discourse.

in order to explain to me the technicalities of the ground oven that had just been set up by the participants of the story. And throughout this parenthetic remark he used only unmarked inflection -nyV which occurs nowhere else in this text.

On the evidence that unmarked inflection -nyV is used in a clearly non-event situation in this context, we can infer from this how a Nyungkul speaker structures his or her narrative into events and non-events by means of this unmarked verbal inflection. In text 12 for instance, lines 5-21 contain predominantly unmarked verb inflection. Here the narrator conveys information about the hero's habits and one of his past exploits, before launching into the story proper (with normal past-tense inflection), which deals with the particular incident of how the hero set a trap for the police. Within the non-event span the narrator briefly changes to past tense inflection in lines 15 and 1 7. These lines deal with the previous encounter between hero and police and thus represent something like an event within a nonevent.

In text 36 variation between past tense and unmarked inflection is somewhat more complicated. The main problem here is that we, as non-Nyungkul outsiders, would probably interpret the utterances containing unmarked inflection as rather belonging to an event sequence. But if we are guided by the presence or absence of unmarked inflection -nyV, we see the plot developing in the following way: lines 1 -4 state that everyone was asleep and that the child had a dream about a crocodile. This information tells us in effect what the story is about and the speaker treats this as event (past tense inflection). Lines 5 and 6, relating that the child did not sleep and cried out, are treated as non-event (unmarked inflection). We already know that the child is having a bad dream and so we may infer that the speaker is now setting the scene for the following event, the interaction between the distressed child and the concerned adults. This, as well as the subsequent narrative up to line 42, is recounted in event form with past tense inflections. In line 43, again perhaps surprisingly because it is in the middle of an action-packed sequence, the speaker switches back to unmarked inflection for non-event. But what is actually happening? The crocodile has got the child and the onlookers are distressed - what more is there to be said? The terrible accident has happened and the following is no more than a fading repetition of the previous happening. (Switching to non-event may perhaps also serve to signal to the audience that this is indeed the end of the child and that there is no hope for a happy turn in the affair.) The narrator then concludes the story with some final event-related remarks (past tense in line 52) to assert the unhappy ending.

H. Hershberger and R. Hershberger's (1 982) interpretation of -nyV as "habitual" would be appropriate for many of its occurrences in the appended texts (e.g. text 1 2, lines 5-14 and text 51, lines 6 1-67), but not for all of them (e.g. text 36, lines 5, 6, 30, 31, 43-47). The term 'unmarked verbal inflection' is preferable as the more general term.

Comparing the various functions of unmarked verbal inflection -nyV one could say that -nyV basically serves to relegate information to a subordinate role. It does so at sentence level by indicating subordinate clauses and it does so at discourse level by identifying nonevents.

5.5.2 Spatial and temporal setting

In Kuku Yalanji, as in many Aboriginal narrative traditions, stories often have centrally to do with the protagonists' movement through time and space. Participants usually move about, sometimes covering considerable distances, with different events happening at

different places. Direction of movement and locations are always clearly defined by the use of locative, ablative, abessive and proper place names or general local descriptions, as for instance throughout text 51.

Spatial setting in Nyungkul can be accomplished by general non-event marking, but another way to set the spatial scene that is common to both Nyungkul and Yalanji is to describe a place in non-past tense as opposed to the past tense narrative. An example of this is found at the very beginning of text 1 2, where the speaker also addresses me personally since he knew that I was familiar with the particular area.

Motion verbs kada-y 'come' and dunga-y 'go' play a significant part in the spatial setting of a narrative. Kada-y is always used with reference to a goal towards which the participant moves, as for instance in text 51, lines 3-4:

• (65 1) 3 Bakarrmuku-muny kada-ny, kada-ny kada-ny kada-ny, Wakooka.Stn-ABL come-PAST come-PAST ... '[They] came from Wakooka Station, came on and on and on,
  • 4 bula yuwulbuwun kada-ny. 3du.NOM(S) Jeannie.River come-PAST they came to the Jeannie River. '

Thus kada-y 'come' serves to indicate both 'transit' (the goal is not yet reached but it is implied) and 'arrival' when the goal is reached.

Dunga-y 'go', on the other hand, may indicate 'departure' as for instance in text 51, line 37:

(652)	Dunga-ny-da	wulbuman	wangkar	dunga-ny	kaya-nda
	go-PAST-TEMP₁	old.woman.ABS(S)	upwards	go-PAST	dog-LOC:pt
	mani-nka.
	get-PURP
	'Off went the old woman (leaving the two young fellows), went up to the dog to get [it].'

As the second occurrence of dunga-ny in the above example implies, dunga-y 'go' may also serve as a term that is unmarked for 'departure', but represents motion from the perspective of the participant, rather than from the perspective of the goal like kada-y 'come'. Consider also text 51, line 5:

(653) Bula kada-ny, bula wawubaja-nga dunga-ny. 3du.NOM(S) come-PAST 3du.NOM(S) river-LOC go-PAST 'They came and they went to the river. '

Here, dunga-ny clearly does not indicate 'departure', since this river is the very goal where they just arrived, but simply 'movement' by the participants.

Thus dunga-y 'go' and kada-y 'come' have the varied discourse function of indicating 'departure', 'transit' and 'arrival'7 as well as changing the perspective of the narrative; kada-y is used from the perspective of the goal and dunga-y from the perspective of the participants.

Narratives usually take place in the past and temporal setting is often established by time words like ngadiku 'a long time ago' or adverbial clauses of time like ngayu karrkayku

⁷ Compare the segmentation of movements into 'departure', 'transit' and 'arrival' in travel and hunting narratives in Ngaanyatjarra (Glass 1 980: 1 23- 1 4 1 ).

'when I was a child', ngayku nganjan warruku 'when my father was a young man', etc. (see §3.8.6.2). However, often a narrative as a whole is not specifically set in a particular time, but verbal inflections are sufficient to indicate that it happened in the past. Temporal setting of episodes within a narrative is accomplished solely by time words, e.g. wunkuny 'next day', or other temporal references such as mudu-waju-ny 'at daybreak', miyil-wujurr-wujurr 'at dusk', etc. (For time words see §3.4.)

5.5.3 Collatera,l: telling what did not happen

Particles kari, yamba and balu have the narrative discourse function of indicating what did not happen out of a range of possibilities, sometimes against expectations. According to Grimes (1 975:65) such "collateral devices" serve to set up alternatives to the actual events in a narrative and the fact that alternatives are mentioned "makes what actually does happen stand out in sharper relief than if it were told without collateral".

The main collateral function of particle kari lies in negating an alternative that is not necessarily stated but is implicit. In §4.3.2 I have shown that particle kari, at the beginning of the second of two conjoined clauses, negates in retrospect the content of the first clause. Kari may also be used initially in an independent clause and we can infer from this that something which could have been expected, but remained unsaid, did not take place; only one of several alternatives happened. Thus the semantic content of particle kari in this context could be paraphrased as 'several alternatives are possible but only the following applies'. Consider an example of two consecutive sentences:

(654) i.	Nganjin-ku-lu 1pl.exc.NOM(A)-EMPH₁-EMPH₂	murruji unable	wukurri-l-wukurri-ny follow-l-REDUP-PAST
	kuyu-lu fish.ABS(S)-EMPH₂	bana-nga water-LOC	janji-ny. swim-PAST
	'We could not follow, the fish was swimming in the water.'
ii.	Kari but	ngamu-ngku-lu mother-ERG:pt-EMPH₂	yalama-ny: "Yubal say-PAST 2du.NOM(A)
	juwal-da digging.stick-INST	dama!" spear-IMP
	'But mother said: "You two spear it with a digging stick!"'

Seeing that they could not follow the fish (which was injured and expected to exhaust itself), these people had two options: leave it alone or try to get it by some other means. Kari implies that one of these options was discarded and the subsequent clause indicates which one was chosen. Another example is in text 51, lines 29-30:

(655) 29	"Ngali	kuwa-muny	kada-ri-ny."
	1du inc.NOM(S)	West-ABL	come-PLS-PAST
	"We two came from the West.""

30 UN gay, kari wangkar-muny ngayu kankada-ny. " INTER but up-ABL I sg.NOM(S) come.REDUP-PAST '''I see, but I came from up there.'"

The second speaker, the old woman with the dog, could have come from several directions, including the one that the first speaker came from. Kari is used to deny this latter possibility

before the woman says where she did come from. In this particular context it is significant that the woman disassociates her own point of departure from that of the two fellows, since she wants to conceal the true nature of her dog which was left behind on the way.

The co-ordination particle yamba (see §4.3.2) qualifies a previous statement, usually describing a state, by adding a contrastive aspect. Hence the use of yamba is akin to the 'adversative' function of kari, for example:

• (656) Walkarr kul-ji kari, black.goanna.ABS(S) rage-COMIT NEG yamba dirra but teeth.ABS(S) germs-mun-ji minya kabu nuka-nya. germs-mun-COMIT meat.ABS(O) rotten.ABS(O) eat-sUB 'The black goanna is not vicious, but [its] teeth are full of germs from eating rotten meat. '
• (657) J ana mala-minya 3pl.NOM(S) expert-meat bubu walu-yindu. land.ABS(S) face-other yala bama, yamba jana-nga like Aborigine but 3pl-poSS.ABS(S) 'They (the Eskimos) are good hunters and fishermen like Aborigines, but their country is different.'

The 'adversative' function of discourse particle balu lies in implying a wrong assumption on the part of the speaker (see §4.9.2 (ii)). Because it is directly speaker related it is used typically in conversation or first person narratives. In third person narratives it may occur in quotations of participants' utterances.

(658) Balu yundu minya dinda-nka, wrong 2sg.NOM(A) meat.ABS(O) roast-PURP bana-nga waju-l. water-LOC cook-NONPAST kari yundu but 2sg.NOM(A) 'I thought you intended to roast the meat, but you are boiling it in water.' (659) Balu nyulu kada-ny-baja.

wrong 3sg.NOM(S) come-PAsT-again 'I thought he was coming back (but apparently he did not).'

5.6 Concluding remarks

This chapter has looked at some aspects of Kuku Y alanji which, although used at sentence level, fulfil their function in the wider context of discourse; indeed, their function cannot be understood at the level of individual sentences. While this chapter has highlighted the discourse significance of certain parts of speech and syntactic devices, such as changing word order and reference strategies, it is by no means an analysis of Kuku Yalanji narrative as a genre. An investigation of Kuku Yalanji genres will require a different approach based on discourse theory and analytical methodology and is outside the scope of this descriptive grammar.

Appendix: texts

The following four texts serve to illustrate not only aspects of Kuku Yalanji grammar and discourse structure but also the unrestricted and active use of this language to communicate all types of subject matter, ranging from everyday events to traditional stories. All four texts, as indeed most of my collection of narratives, were told spontaneously.

The most prolific narrators as well as capable editors of their own texts were the Kuku Nyungkul speakers Charlie Tayley and Bobby Roberts, both in their fifties. Born in Cooktown, Charlie Tayley was taken on as a station hand in his early teens by the owner of Greenhill Station. He worked on a number of stations in the south-eastern part of the Cape York Peninsula, droving cattle from as far north as Wakooka Station on Princess Charlotte Bay to as far south as Bowen, before settling at Wujal Wujal with his family to allow his children a regular school education. At the time of my fieldwork in Wujal Wujal he was the head stockman of the mission and was acknowledged as a good storyteller by the community. He passed away in the mid 1 980s. During his days as a drover, in the company of older Kuku Nyungkul, Kuku Yalanji and also Guugu Yimidhirr speakers, he acquired a large repertoire of mythical stories (text 51 is one example) and tales from the time of the Palmer River goldrush ( 1 873 to about 1 885). He also had many of his own anecdotes to tell from his childhood and droving times. His source of Palmer River goldrush stories was an old man, Billy Charcoal, who lived at Laura Station and would entertain the young stockmen with eyewitness accounts of fighting, guerilla warfare and trickery on the part of the Aborigines, who fought a bitter but losing battle for their lives against the marauding European and Chinese intruders.} Text 12 is one of the tamer stories from this time.

Bobby Roberts spent his childhood and youth on Palm Island, where his parents had been transported when he was only a small boy. He grew up speaking English, the lingua franca among Aborigines from different tribes who lived together on the prison island and attended school. From his parents he acquired only a basic knowledge of Kuku Nyungkul. On returning to the Bloomfield River in his early twenties he made a conscious effort to learn his tribal language properly from the older people, as well as acquiring information on bushcraft, native food and traditional customs. He took it upon himself to devise lessons for me on these subjects, providing me first of all with the necessary vocabulary and then illustrating his topics with narratives. Text 36 was part of one such lesson, dealing with the significance of dreams. Bobby Roberts passed away in the mid 1 980s.

Ivy Walker, a Kuku Yalanji speaker in her mid-forties, had lived all her life in her tribal country. She had a thorough knowledge of Yalanji myths and folklore, but her favoured topics of conversation were family matters and events from her own life, particularly

See also Anderson and Mitchell (1981 :2 1-37) and Reynolds (198 1 : 1 54).

222 Appendix

camping trips that her family had undertaken together with Chris Anderson, an anthropologist from the University of Queensland, whom she called "younger brother". The excerpts in text 3 are taken from a very long narrative recounting a week-long trip to Shiptons Flat, her husband Johnny Walker's (Kuku NyungkuI) traditional territory. The text was transcribed and edited to some extent with the help of Queenie Dick in 1 979. Owing to Ivy Walker's sudden death early in 1981 no final checking and editing of her texts was possible.

Text 51: Nyungkul

Charlie Tayley

Why the dingo always turns around before running away

1. Warru-bulal jambul, bula jilba-dunga-ny yg.man-pair.ABs(s) two.ABS(S) 3du.NOM(S) walk-go-PAST

dunga-ny dunga-ny, gO-PAST gO-PAST 'Two young men, they went walkabout, went on and on, '

2. bula kuwa-muny kada-ny, wanja-muny bakarrmuku. 3du.NOM(S) West-ABL come-PAST somewhere-ABL Wakooka.Stn 'they came from the West, from somewhere (around) Wakooka Station.'

• 3. Bakarrmuku-muny kada-ny, kada-ny kada-ny kada-ny, Wakooka.Stn-ABL come-PAST come-PAST ... '[fhey] came from Wakooka Station, came on and on and on, '
• 4. bula yuwulbuwun kada-ny. 3du.NOM(S) Jeannie.River come-PAST 'they came to the Jeannie River. '
• 5. Bula kada-ny, bula wawubaja-nga dunga-ny. 3du.NOM(S) come-PAST 3du.NOM(S) river-LOC gO-PAST 'They came and they went to the river. '
• 6. Bula wawubaja-nga dunga-ny, bula 3du.NOM(S) river-LOC gO-PAST 3du.NOM(A)

mayi kabal warrmba-bunga-ny. fruit.ABs(o) Leichardttree.ABS(O) uncovered-CAUS-PAST 'They went to the river and they found some Leichardt fruit.'

• 7. Kabal mayi warrmba-bunga-ny. Leichardttree.ABS(O) fruit.ABS(O) uncovered-CA US-PAST 'Leichardt fruit [they] found.'
• 8. "Yah, mayi ngali-nga kabal. " INTER fruit.ABS Idu.inc-POSS.ABS Leichardttree.ABS "'Hey, there is some Leichardt fruit for us.'"

l

• 9. Bula 3du.NOM(A) nuka-ny nuka-ny nuka-ny nuka-ny, eat-PAST ... 'They ate and ate and ate and ate, ,
• 1 0. yinduy-nju yalama-ny: "N gali daka-l? one.of-ERG(A) say-PAsT ldu inc.NOM(S) climb-NONPAST '[and] one of them said: "Shall we climb?'"
• 1 1 . N gali juku-ngu daka-l. " Idu.inc.NOM(S) tree-LOC climb-NONPAST "'We will climb on the tree.'"
• 1 2. Bula daka-ny. 3du.NOM(S) climb-PAST 'They climbed.'
• 1 3. Daka-ny, juku-muny mani-l-mani-ny mayi nuka-l-nuka-ny. climb-PAST tree-ABL get-l-REDUP-PAST fruit.ABS(O) eat-l-REDUP-PAST '[They] climbed up, kept taking fruit from the tree [and] eating it.'
• 1 4. Bula yarra wangkar nyaji-ny. "Yah! " 3du.NOM(A) direction upwards see-PAST INTER 'They looked up that way. "Hey!'"
• 1 5. Wulbuman wangkar-mundu kaya-nji kankada-ny. old.woman.ABS(S) up-ABESS dog-COMIT:pt come.REDUP-PAST 'An old woman was coming from up there with a dog.'
• 1 6. Wulbuman-anga kaya yalbay. old.woman-POSS.ABS(S) dog.ABS(S) big 'The old woman's dog was big. '
• 1 7. Kada-ny kada-ny kada-ny kada-ny, come-PAST ... '[The woman and her dog] came on and on and on and on, ,
• 1 8. wangkarwangkar kaya bawa-ny. up.there dog.ABS(O) leave-PAST '[and she] left the dog up there.'
• 1 9. Wangkarwangkar bawa-ny, nyulu badabada dunga-ny. up.there leave-PAST 3sg.NOM(S) down.there go-PAST '[Having] left [the dog] up there, she went on down.'
• 20. Badabada dunga-nya, wawubaja-nga kada-ny, down. there go-SUB river-LOC come-PAST 'Going on down [she] came to the river, '
• 21. kabal yinya-ku nuka-l-nuka-ny bula. Leichard.tree that-EMPHI eat-l-REDUP-PAST 3du.NOM(A) '[at the] very same Leichard tree [where] the two [men] were eating.'
• 22. Nyulu nuka-ny nuka-ny nuka-ny, 3sg.NOM(A) eat-PAST ... 'She ate and ate and ate, '
• 224 Appendix
• 23. bula wangkarwangkar-mundu kuni-ny nyunguniny 3du.NOM(A) up.there-ABESS hit-PAST 3sg.ACC(O)

mayi-bu wawumu-bu dukul-baja-ku. fruit-INST half.ripe-INST head.ABs(O)-DEGI-EMPHl 'and the two hit her from up there with half ripe fruit right on the head. '

• 24. Nyulu wulman-angka waba-ny. "Yah!" 3sg.NOM(A) old.woman-ERG:pt look.up-PAST INTER 'The old woman looked up. "Hey!'" (Note, waba-l is formally a transitive verb, but does not require a transitive object.)
• 25. Bulaniny nyaji-ny wangkarwangkar bundanda-nya jalban-ba. 3sg.ACC(O) see-PAST up.there sit.REDUP-SUB tree top-LOC '[She] saw the two sitting up there in the tree top. '
• 26. "Yah, kada-y kada-y! " INTER come-IMP come-IMP "'Hey, come [down], come!'"
• 27. Bula kada-ny. 3du.NOM(S) come-PAST 'The two came.'
• 28. Wulbuman-angka balka-ny: ''Yuba I wanja-muny kada-ny?" old.woman-ERG:pt(A) speak-PAsT 2du.NOM(S) where-ABL come-PAST 'The old woman spoke: "Where did you two come from?'"
• 29. "Ngali kuwa-muny kada-ri-ny. " I du.exc.NOM(S) West-ABL come-PLS-PAST "'We two came from the West." ,
• 30. "N gay, kari wangkar-muny ngayu kankada-ny. " INTER but up-ABL I sg.NOM(S) come.REDUP-PAST "'I see, but I came from up there.'"
• 31. "N gay. " INTER "'Oh, all right." ,
• 32. Balka-wa-ny balka-wa-ny balka-wa-ny bula-nji. speak-REcIP-PAST ... 3du-COMIT:pt '[She] talked and talked and talked with the two. ' (i.e. they all had a long chat)
• 33. "Bunda-y", yalama-ny, "yubal, yubal yalaymba bunda-y! sit-IMP say-PAST 2du.NOM(S) 2du.NOM(S) here sit-IMP "'Stay (lit. sit)", [she] said, "you two, you two stay here!'
• 34. N gayu kayu karrkay bawa-ny wangkarwangkar. I sg.NOM(A) dog.ABS(O) little.ABs(o) leave-PAST up.there 'I left a little dog up there. '
• 35. Wawu-wula-ny nyulu, ngayu nyunguniny bawa-ny. " breath-die-PAST 3sg.NOM(S) I sg.NOM(A) 3sg.ACC(O) leave-PAsT 'It was exhausted and I left it." ,
• 36. Bula nyaji-ny jakalbaku. 3du.NOM(A) see-PAST before 'The two had seen [it] before.' (i.e. they knew that it was not a little dog at all and that the old woman wanted to deceive them.)
• 37. Dunga-ny-da wulbuman gO-PAST -TEMPI old. woman.ABS(S) wangkar dunga-ny upwards gO-PAST

kaya-nda, mani-nka. dog-LOC:pt get-PURP 'Off went the old woman, went up to the dog to get [it].'

• 38. Bula warru-bulal jalama-ny-kuda, 3du.NOM(S) yg.man-pair.ABs(s) jump-PAsT-TEMP2 'In the meantime the two young fellows took off, '
• 39. jalama-ny yarra-kuda bayan-ba kuwa warri-ny. jump-PAsT direction-TEMP2 camp-LOC West run-PAST '[they] took off in that direction and ran towards [their] camp in the West.'
• 40. Wulbuman dunga-ny, kaya mani-ny. old.woman-NOM(S) go-PAST dog.ABS(O) get-PAST 'The old woman went and got the dog.'
• 41. Kaya mambarri-ny bulaniny. dog.ABS(O) order/force-PAST 3du.ACC(O) '[She] set the dog on the two. '
• 42. Kaya badabada dunga-ny, yambakari. dog.ABS(S) down. there go-PAST nothing 'The dog went down there, nothing. (The two were gone.)'
• 43. Kaya-ngka wukurri-ny-da bulaniny. dog-ERG:pt(A) follow-PAST-TEMPI 3du.ACC(O) 'Then the dog followed the two. '
• 44. Nyulu wulbuman kudamundu kankada-ny. 3sg.NOM(S) old.woman.ABS(S) behind come.REDUP-PAST 'The old woman was coming behind.'
• 45. Kaya-ngka wukurri-ny wukurri-ny wukurri-ny. dog-ERG:pt(A) follOW-PAST ... 'The dog followed and followed and followed [them].'
• 46. Bula kuwa dunga-ny-arrku, kuwa dunga-ny-arrku, 3du.NOM(S) West go-PAST-EMPH4 West go-PAST-EMPH4 wuburr-bu kada-ny. mountain. top-LOC come-PAST 'The two went right on to the West, went right on to the West and came to the top of the mountain. '
• 226 Appendix
• 47. Kangkay-mba kada-ny. Jones'.Gap-Loc come-PAST '[They] came to Jones' Gap.'
• 48. Bula bundanda-ny yinyaymba, yindu 3du.NOM(S) sit.REDUP-PAST there one.of.ABS(s)

kungkar-mundu yindu jibarr-mundu. North-ABESS one.of.ABS(S) South-ABESS 'They stayed there, one on the northern side, the other on the southern side.' (i.e. one looking down from one side, the other looking down from the other side)

49. Kaya wanarri-ny, kaya warri-ny warri-ny warri-ny, dog.ABs(S) run.REDUP-PAST dog.ABS(S) run-PAST ... 'The dog ran on, the dog ran and ran and ran, ,

• 50. bula dama-ny. 3du.NOM(A) spear-PAST '[and] the two speared it. '
• 51. Dama-ny, kuni-ny, kanku nyanda-ny. spear-PAST kill-PAST half cut-PAST '[They] speared [it], killed [it] and cut [it] in half. '
• 52. "Yunu kanku, ngayku kanku." 2sg.POSS half I sg.POSS half "'One half for you and one half for me.'"
• 53. Dakil nyungu yinduy-mbu. front.leg.ABs 3sg.POSs one.of-POss 'One of them [got] the front legs [half].' (Note that yindu has a suppletive y-final root in inflected form. Inflections are always from the 'neutral' set.)
• -------_._---------------------------------- 54. Wundi-ny-da bula wundi-ny wuburr-bu take-PAST-TEMPI 3du.NOM(A) take-PAST mountain.top-LOC wangkarwangkar, wanjabu Wakuka. up. there somewhere Wakooka.Station 'Then they took [the dog pieces] and took them up there to the mountain top somewhere around Wakooka Station. '
• 55. Bama wubul yinyaymba bundanda-ri-ny. Aborigine.ABS(s) many.ABS(S) there sit.REDUP-PLS-PAST 'Many people stayed there. '
• 56. lana jurrkiji-ny minya-ji, 3pl.NOM(S) shift.camp-PAsT meat-COMIT 'They shifted camp with the meat, '
• 57. minya jurrki-ny jana, badabada dunga-ny. meat.ABS(O) move-PAST 3pl.NOM(A) down.there gO-PAST 'they moved the meat and went down there.'

58.	Badabada dunga-ny. down.there go-PAST '[They] went down there.'
59.	Wawubaja-nga kada-ny, dinda-ny minya, kurrma-nga river-LOC come-PAST roast-PAST meat.ABS(O) ground.oven-LOC dinda-ny. roast-PAST '[They] came to the river and roasted the meat, roasted [it] in a ground oven.'
60.	Kurrma-nga dinda-ny. ground.oven-LOC roast-PAST '[They] roasted [the meat] in a ground oven.'
	(The following is an aside explanation on how people used to roast meat in a ground oven; note the unmarked verbal inflection from line 61 to line 67.)
61.	Ngadiku dubal nyanda-nya, long.ago bark.ABS(O) chop-UM 'In the old days [they] chopped bark [off a tree],'
62.	minya yaka-ji-nya yaka-ji-nya, meat.ABS(S) cut.up-ITR-UM cut.up-ITR-UM 'meat was cut up into pieces,' (agentless Passive)
63.	bana nyangarri-nya dubal-ba, water.ABS(O) pour-UM bark-LOC '[they] poured water into the bark,'
64.	kurrma waju-nyu, ground.oven.ABS(O) burn-UM 'lit the ground oven,' (i.e. the fire)
65.	kurrma-nga yijarri-nya minya, ground.oven-LOC put-UM meat.ABS(O) 'put the meat in the ground oven,'
66.	buyku bundanda-nya. paperbark.ABS(S) sit.REDUP-UM 'and paperbark [leaves] sat on top.' (i.e. covered the meat)
67.	Bubu-ngu bundanda-nya, bawa-nya. ground-LOC sit.REDUP-UM leave-UM '(The oven) stayed in the ground and was left alone.'
	(End of parenthetic remark, note return to PAST inflection in the next line.)
68.	Jana bunda-ny bunda-ny bunda-ny, 3pl.NOM(S) sit-PAST ... 'They sat and sat and sat,'
69.	minya waju-ny, walnga-ny. meat.ABS(O) cook-PAST take.out-PAST

'[when] the meat was cooked, [they] took it out.'

• 228 Appendix
• 70. Walnga-ny jana minya take.out-PAST 3pl.NOM(A) meat.ABS(O) 'They took the meat out and then ate it. ' nuka-l-nuka-ny-da. eat-l-REDUP-PAST-TEMPI
• 71. Nuka-ny nuka-ny nuka-ny nuka-ny. eat-PAST ... '[They] ate and ate and ate and ate. '
• 72. Nyulu wulman yanya-bi 3sg.NOM(S) old.man.ABS(S) this-EMPH3 'There was this old man sitting (there).' bundanda-ny. sit.REDUP-PAST
• 73. ' Nyulu wulman wabarr-ba 3sg.NOM(S) old.man.ABs(s) shade-LOC 'The old man was sitting in the shade.' bundanda-ny. sit.REDUP-PAST
• 74. Yinduy-nju yalama-ny: one.of-ERG(A) say-PAST 'One of them said: '

75. "Yah, yundu wulman-anda babaji-ka-rr INTER 2sg.NOM(S) old.man-LOc:pt ask.about-IMP-TEMP3

nyulu minya-ka wawu." 3sg.NOM(S) meat-DAT want "'Hey, do ask the old man [if] he wants some meat!'"

• 76. Yindu dunga-ny nyungu-ndu babaji-ny: one.of.ABS(S) gO-PAST 3sg-LOC:pt ask.about-PAST 'One of them went and asked him:'
• 77 . "Yah, wulman, yundu minya-ka wawu ?" INTER old.man 2sg.NOM(S) meat-DAT want "'Hey, old man, do you want some meat?'"
• 78. "Mh mh" dukul wuri-nya. head.ABS(O) move.back.and.forth-SUB "'Mh rnh" [said the old man] nodding his head. '
• 79. "Wanyurri-ngku minya-ka, yundu bangkarr-ka wawu ?" what-DAT meat-DAT 2sg.NOM(S) steak-DAT want "'What [sort of] meat; do you want steak?'"
• 80. "Kari. " INTER "'No.'"
• 81. "Yundu jiba-ka wawu?" 2sg.NOM(S) liver-DAT want "'Do you want liver?'"
• 82. "Kari. " INTER '"No.'''
• 83. "Dakil-ka yundu wawu, malbin?" front.leg-DAT 2sg.NOM(S) want hind.leg.ABs "'Do you want a front leg, [or] a hind leg?'"
• 84. "Kari. " INTER '"No.'' '
• 85. Wawu kari nyulu. want NEG 3sg.NOM(S) 'He did not want [any of these].'

86. Yinduy-nju yalama-ny: "Yundu dukul-ku wawu?" one.of-ERG(A) say-PAST 2sg.NOM(S) head-DAT want 'The other one said: "Do you want the head?'"

• 87. "Yuwu!" INTER "'Yes !'"
• 88. Dukul nyungu daji-ny. head.ABS(O) 3sg.POSS.ABS(O) give-PAsT '[The other one] gave him the head. '
• 89. Dukul mani-ny, nyulu bada yilki-nga dunga-ny. head.ABs(o) take-PAST 3sg.NOM(S) downwards creek-LOC gO-PAST 'He took the head and went down to the creek. '
• 90. Bayan-muny nyulu bada yilki-nga dunga-ny. camp-ABL 3sg.NOM(S) downwards creek-LOC go-PAsT 'He went from the camp down to the creek.'
• 91. Bada yilki-nga dunga-ny, downwards creek-LOC gO-PAST '[He] when down to the creek,'
• 92. nyulu yinya bajibay 3sg.NOM(A) that.ABs(o) bone.ABS(O) 'he gathered up those bones,' muja-ny, gather-PAsT
• 93. junumunda-ny-da. join-PAST-TEMPI 'and then joined them together.'
• 94. lunumunda-ny junumunda-ny junumunda-ny. join-PAST ... '[He] kept joining and joining and joining. '

95. Dakil balka-ny, dakil balka-ny, malbin, front.leg.ABS(O) make-PAST front.leg.ABS(O) make-PAST hind.leg.ABs(o) malbin, muku, dukul, milka, milka, biji, back.ABS(O) head.ABs(o) ear.ABS(O) ... tail.ABS(O) '[He] made a front leg, made [the other] front leg, a hind leg and [the other] hind leg, back, head, ears, tail, '

• 230 Appendix
• 96. nyulu yalama-ny: 3sg.NOM(A) say-PAST "Yundu warndi-y! " 2sg.NOM(S) get.up-IMP '[and] he said: "Get up!'"
• 97. Nyulu warndi-ny 3sg.NOM(S) get.up-PAST 'It got up, the dog [did].' kaya-Ia. dog.ABS(S)-EMPH 2
• 98. Nyulu kaya warndi-ny. 3sg.NOM(S) dog.ABs(S) get.up-PAST 'The dog got up, '
• 99. "Yundu walu-wukurri-ka!" 2sg.NOM(A) face-follow-IMP '[and the old man said:] "Turn around!'"
• 1 00. Nyulu walu-wukurri-ny, dunga-ny. 3sg.NOM(A) face-follow-PAST go-PAST 'It turned around and went [off].'
• 101 . "Yundu warri-y, yundu kurrka-waba-ji-ka!" 2sg.NOM(S) run-IMP 2sg.NOM(S) behind-Iook.up-ITR-IMP '[The old man said:] "Run, and look behind you !'"
• 1 02. Nyulu warri-ny, kurrka-waba-ji-ny, 3sg.NOM(S) run-PAST behind-look up-ITR-PAST 'It ran, and looked behind, '
• 1 03. warri-ny-baja, kurrka-waba-ji-ny. run-PAsT-again behind-Iook.up-ITR-PAST 'ran again, and looked behind. '
• 1 04. Nyulu yajarn-Jl-ny nyulu-rrku: 3sg.NOM(S) laugh-ITR-PAST 3sg-EMPH4 'He [the old man] laughed to himself:'
• 1 05. "Ha ha ha, yala-ku yundu bunda-y!, ' thiS-ADV 2sg.NOM(S) sit-IMP "'Ha ha ha, you stay like this !'"
• 1 06. Nyulu kaya, yala-ku nyulu 3sg.NOM dog.ABS thiS-ADV 3sg.NOM(S) 'The dog, he stayed like this.' bunda-ny. sit-PAST

"That's the one, dingo."