Papers in Australian Linguistics No. 1

PACI FIC LINGUISTICS

Edi tor: S. A. Wurm Associate Edltors: D.C. Lqycock, C.L. Voorhoeve

SERI ES A - OCCASI ON AL PAPERS

No . 10

PAPERS IN AUSTRALIAN LINGUISTICS No.1

D . an d K. GLASGOW

JEAN F. KIRTON

W.J. OATES

S.A . and E.G. SOMMER

CANBERRA 1967

THE AUSTRALIAN NATIONAL UNIVERSITY

Glasgow, D., Glasgow, K., Kirton, J., Oates, W., Sommer, S.A. and Sommer, E.G. editors. Papers in Australian Linguistics No. 1. A-10, vi + 65 pages. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-A10.cover ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. PACIFIC LINGUISTICS is published by the Linguistic Circle of Canberra and consists of four series:

SERIES A - OCCASIONAL PAPERS SERIES B - MONOGRAPHS SERIES C - BOOKS BULLET INS.

All correspondence concerning PACIFIC LINGUISTICS, including orders and subscriptions, should be addressed to:

The Sec ret ary, "Pac ific Linguistics" , Department of Linguistics, Schoo l of Pac ific Studies, The Australian Nat ional Un ivers ity,

Canberra, A. C. T. , Australia.

The copyright in the material of this publication is vested in the authors.

The editors are indebted to the Australian National Uni versity for help in the production of this series.

This publication was made possible by a grant from the Hunter lliuglas Fund. TABLE OF CONTENTS

Pa�e THE PHONEMES OF BU RERA, by David and Kathleen Gl asgow 1

1. In troduction 1

2. The Phonemes 1

3. Consonantal Contrasts 2

4. Co nsonantal Variants 3

5. Vo calic Co ntrasts 4

6. Vo calic Variants 4

7. Stress 7

8. Distribu tion of Phonemes 7

9. The Sy llable 8

10. Distribution of Sy llabl es 8

11. Interpretation 9

12. Proposed Orthography 10

No tes 14

ANYULA PHONOLOGY, by Jean F. Kirton 15 O. Introduction I5

1. Phonem es 15 1.1. Consonantal De scription 15 1. 11. Consonantal Contrast 16 1. 12. Consonantal Variants 18 1. 2. Consonant Distribution 19 1. 3. Vowel Descrip tion 21 1.31. Vowel Contrast 21 1. 32. Vo calic Variants 22 1. 4. Vowel Distribution 23

2. Units Larger than the Phoneme 23

iii iv

Paf1e

2. 1 Phonological Syllable 23

2. 2. Phonological Wo rd 24

27 No tes

Bibl iof1raphy 28

SYLLABLE PATTERNING AND PHONETICALLY COMPLEX CON SONANTS

IN SOME AU STRALIAN LANGUAGES, by W. J. Oates 29

O. Introduction 29

1. Predomin an t Syllable Patterns in Au stral ian Languages 32

2. Sy llable Co nsonan t/Contoid Fi llers 32

3. Interpretation of Homorganic Nasal- Stop Sequences 33 3. 1. Problems of Interpretat ion in some Languages (such as Gugu-Yalanji) 33 3. 2. Interpretat ion in Gugu- Yalanj i of the NS Syllable Unit 34 3. 3. Binary Nature of the Homorganic Nasal-Stop Phoneme s in Gugu-Yalanj i 34 3. 4. Phonemic Status of Component Parts of the NS Unit in Gugu-Yalanj i 35

4. Interpretation of Retrofl exed Co nto ids 36 4. 1. Binary Nature of Retro fl exion in Tiwi 36

5. Obj ections to An alysis 37

6. Defence of Analysis 38 6. 1. CV Patterning 38 6. 2. Patterning of /r/ in Non-retroflexed Stop and Nasal Sequences 38 6.3. Morpheme Boundaries 39 6 .4. The Stability of the /r/ Phoneme 39 6 . 5. Retroflexion of Vocoids 40 6.6. Rarity of Wo rd Initial Retro flexion 41

No tes 42

Charts A, B, C, D, E pages 44-49

Bibl i of1rap hy 50 v

KUNJ EN PRONOUNS AN D KINSHIP, by B. A. and E. G. Sommer 53

O. Introdu ction 53

1. Pronouns 53

2. So ci al o r g an i sat ion: Kinship 55

3. Kinsh ip Terms 55

4. Co r reI at ion 56

5. Residual Fo rms 57

No tes 58

ACknowLedf,ments 59 Glasgow, D., Glasgow, K., Kirton, J., Oates, W., Sommer, S.A. and Sommer, E.G. editors. Papers in Australian Linguistics No. 1. A-10, vi + 65 pages. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-A10.cover ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. THE PHONEMES OF BURERA

DAVID and KATHLEEN GLASGOW

1. INTRODUCTION The purpose of this paper is to present the phonemes of the Burera language 1, and to propose a practical orthography for vernacular literacy wh ich will allow the smoothe st tr an sition into English. Burera is spoken by approximately 350 people of the Blyth River area in Ar nhem Land. There are two main dialects dis tinguished by the way they say that one, kun-narda or kun-nartpa respectively. The material for th is paper was gathered from April 1962 to September 1964. at the Maningri da Settlement on the mouth of the Liverpool River, where most of the Burera speakers have settled. The author s are grateful to Miss Eunice Pike of the Summer Institute of Linguistics, who was a most valuable consultant in the final analys is and wr ite-up of th is paper . They also wish to acknowledge the helpfulness of material shared by Lester Hiatt, Lecturer in Anthropology at Uni versity of Sydney.

2. THE PHONEMES The eighteen segmental phonemes are represented in the following charts . CONSONANTS: Bi labi al A lveo Al veolar Ve lar Denta l

St ops p t k

Na s als m n

Flap

Late ra l 1

Non-syllab ic Voco ids w y r

Glasgow, D. and Glasgow, K. "The Phonemes of Burera". In Glasgow, D., Glasgow, K., Kirton, J., Oates, W., Sommer, S.A. and Sommer, E.G. editors, Papers in Australian Linguistics No. 1. A-10:1-14. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-A10.1 ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. 2

VOWELS: Fr ont Cen tral Back Rounded

Low a

There is a suprasegmental phoneme of stress.

3. CONSONANTAL CONTRASTS The stops Ip , t, t, kl contrast as follows: 'prelapila table munku 'prekka lon� stick trel �round IJu'trekka [w iLL ret,urn 'trelippre chiLd IJu'trertre [ am stron� , krelrernre ear kuna ' kreppa dawn

'tara�ap La st one an 'tampa"t �ood hunte r an' marpat skeLet on , IJanpak a rm Let

Ipl contrasts with /w/ as in 'paIJarri name of subsection and 'waIJara spirit, a'pun� hit him and a'wun� �ave to him, a'prenre he arrived and a'wrenre he spoke,

It I contrast s with Iyl as in '"tiri"t"ti water LiLy and 'y�r i"ttiIJa name of mo iety, ku'treritrer itiIJa pours it and ku' yrer iyrer itiIJa it fLows , 'kulat �rass LiLy and a'lay hey! (m asc.J.

It I contrasts with Irl as in 'putippre they two and 'pu rippre they pLuraL, tinku' furta catfish and pur ' rak cont inu in�, 'trel ippre chiLd and 'rreppura foot .

Irl and /11 contrast as in 'para buttoc ks and 'pala roof, 'kurr� here and an 'k�rlQ dry , baLd . They contrast with Irl as in 'p�ri"ta upLand �rass and 'p�rittal rice �rass, '"tarpar cypres s pine and 'watpar type of oak, 'fi�lkkiya you tease and 'fbrkkiya you do habituaLLy , 'lama shoveL spear head and 'rama� fLuff; 'tarakkara Lift it uP!, 'talakkar road, and 'tararJa sand.

The nasals 1m, n, n, IJI contrast as follows : 3

'm<:>l<:> a�a in 'lama shovel spear head 'n<:>lkkiya you tease 'ana son, father 'n<:>kk<:>r<:> small shark 'lanappi�a cypress pine '�<:>rl<:> didjeridoo '�ala�a sister's son

'mamam mother 's mother 'pulan name of subsectton 'raman ! luff 'kalama� axe

The non-syllab ic vocoids Iy, w, rl contrast as follows:

'yamu can? a'ya:Hni;;i wi II throw him 'wamut name of subsection a'warli;;i;;i he wi II �o up 'raman fluff a'rakkiya he sits down

a' lay hey! (ma s c . ) 'watpar type oak �aw yes

4 . CONSONANTAL VARIANTS The stops /p , i;;, k/ have voiced var iants which al ternate with voicelessn ss wo d initially as in /p<:>l/ <:>l/ � � _ !p b<:>IJ fire, /'i;;urai ['i;;urA/'�urAJ paper, /'kuni;;<:>�/ ['kun�<:>�/ 'gun�<:>�J tree . These voiced variants occur word medially except in stop clusters2 . For example /apuri'i;;rekkarre/ [abur.i'�kAr8J they returned , /an'kuna/ [an'gunAJ this one. The voiced variant of /t/ alternates with voi celessness word initially, and med ially preceding Ip/ as in /'titpurkl ['�Itpurk/'�IdpurkJ axe; and occurs me dially preced ing the stressed vowe l and following nasals as in /�u'trelipprel [�u'drelip8J I am small, I'pantre/ ['pand8J le�. The alveo ar consonants It, n, r, II have retroflexed variants. [�, �, �J occur word initially and in con sonant clusters following /r/ as in I'tiyamal ['tiyamAJ type e shel lfish, Ii;;inku'purtal [i;; nku'burt. AJ lar� catfish, I . I 'n<:>rnt<:>/ ['�<:>rl?-�<:>J mud whelk shel l fish, /'lama/ ['�amAJ shovel spea r head, I'w<:>rbl ['w<:>r�<:>J siblin�. [fJ occurs word medially following /r/ as in /'�arrappa/ ['�arfapAJ a lone. 4

Ir/ has a trilled variant which tends to occur word fi nally and in consonant clusters except following Irl as in I Iy';)rl [y';):fJ ra tn, / 'yrermpre/ ['ysrmbsJ husband , wife S b rot he r, /' �anaral [, �anarAJ or shor tened form [, �a nrAJ rock.

5. VOCALIC CONTRASTS There are five vowels i.i, re, a, ';), ul wh ich contrast on stressed syllab le as follows : '�iri��i LiLy 'kippre just now 'treri-tpri�i pour kuni'kreppre dawn '�ari��a carve 'kappal erassLand '� ';):Hfi�ur';) h ieh ero und ku'k';)ppur';) pers onaL pos- sesston '�ura paper ku'kuppur';) buiLds it kana'pinpin bat �u'wirippi I wiLL sPiLL a'pEnre he arrived a'wrefremiya he tS hunery 'pantre Lee a'waritti�a he eoes up a'p';)n';) he went he watches cor roboree a 'pun';) he hit him 'wuru man

, pilimura Le e c h a'rimu wiLL hoLd him 'prelapila tabLe a'rremu wiLL hammer him 'pala house a' ranre spea red him p';)l fire a'r';)�a he thirsted, burned 'pulafi name of subsection a'rumu wiLL brea k him

6 VOCALIC VARIANTS The area of vocalic var iat ion for each vowel, vocalic glides exc luded , and the overlap between vowe l areas are shown on the following chart. 5

Fron t Cen tral Back

J u/ / i /

/ ';)/ Low

/ref / a/

The following chart shows the var iants of each vowel when in stressed syllable in respect to its immediate environ ments . A blank space has been le ft where no example has been recorded. Preceding Following Con sonan t Con sonan t .i a U p p re.l a U alveodental/y velar a U alveodental/y w reI U alveodental/y other reI ';)/ U ';r I- alveodental/y a1 ';)/ ';r U f other e/e1 a ';)/ ';r U w alveodental/y I a1 ';)/ ';r U w velar I/e a ';).l/o oUt U w other I a ';) / ';r U velar p e/e1 a ,;)/,;).l o/u velar alveodental I a1 ,;)/,;).l u/.... velar y .i a1 ,;)/,;).l u/.... velar 1 I a ';)/';). l U velar velar 8/e a ';)/';)U oU/u velar other I a ';)/';).l u / other alveodental y i a1 ';)/ G u/ .... other velar I a ';)/';)u oUt u other other I a ';)/ ';r u word initial .i a word final I/e a u 6

In unstressed syllables vocalic variant s tend to central ize toward [8J and are otherwise conditioned by consonantal environment s as in stressed syllables. Word final unstres sed lal is [AJ as in I'palal ['palAJ house. Examples of vocalic variants are as follows : The norm of I il is [IJ as in I ·t iIJkal [·tIIJgAJ pandanus fruit. l'IJ inipprel ['IJIInIpeJ you (s .), lan'p irlirlrel [an' bIr lIr leJ 'h is liver . Other variant s of Iii are as in la' pippitiIJal [a.bipIC;IIJAJ he blows , la·rimuIJal [a.remuIJAI a·re1muIJAJ holds him, I 'wikipprel ['wIgIpel 'wegIpeJ toeet her. I'kipprel ['kepe/'ke 1peJ just now, I'kiyal [ .k iYAJ eee. l'IJikkal [ ' r;)ekA/'IJekAJ no. j·mittiyaIJI ['mitIyaIJJ ship. I',i [ 'iYAJ qu' es tion, la·nil [a ' nIyal/a·n eJ he wi ll sit. The norm of lrel is [re�J as in lakrerrek 'pawul [agre�rre�k 'bawuJ he wiII dodee. l a·mreIJrel [a'mre�IJeJ eot him. I 'prela pilal [ ' pre�labI1AJ table. Other variant s of lrel are as in / ' �rekkal [ ·trekAJ eo back!, I ·yrekkal [ 'yrekAJ ea rly dry seas on, la·trew iti/ [a ' c;ewI(jjIJ he wil l eet up to eo huntine.

I t re 1I r tel J e r 0 u n d , I IJU • r refiti I [ IJ u• r re 1fiC; I J I w ill put weie ht on. stamp, I ·rreIJarreIJal [.rreIJarreIJAJ day. l ie ht. /a·wrettawrettiIJal [a.weltawe1tIIJAJ he searc hes, / ·wreIJka/ ['wreIJgAJ speak!, I·wreppal ['wrepAJ wa sh!, /kuna·krepprel [kuna ' grepAJ it dawns, la'IJret tiIJal [a' IJe1tIIJAJ it eives birt h, I'kreyla/ ['ke1ylAJ name of subsection, I'krelremrel [ 'kre�lre�meJ ea r, IIJu' IJrekkal bJu ' IJrekAJ I wiII re s t, I· mret tak/ [·meltakJ feather fa n, IIJu'preypu/ [IJu'be1ybuJ I wil l pass, / 'reyw'Ol ['e1yw'O J ques t ion. The norm of /al is raJ as in I ·kala/ [·kalAJ neeat ive. I ·wakarpul ['wagarbuJ shou lder, I'palmarkl ['palmarkJ wind. I ·amal ['amAJ mo ther, lana 'kal [ana.gaJ he wi ll brine. Other variants of lal are as in /·tawul [·tre�wuJ throat. /·yawuHfi/ ['yre�wurifiJ sine le ma n, I·talrel [·tre�leJ sister. /·yalppul [·yredpul cook it!, Ip i'rat til [pI'ra ltIJ rice. Ipa 'rayre/ [pa'ra1yeJ la.ter, Iku 'wat t inal [ku 'waltInAJ it bro ke. /'waykkun/ [·wa1ykunJ above. /ku'ka�ti/ Lku.galtIl it wi ll dry up. /'kayut/ ['ka1yutJ shovel spear. /an'pattira/ [�n'baltIreJ fierce. The norm of /'0/ is ['O/oJ as in / y'Or/ [y'Or/ YGFJ rain. /mu 'r'O�a/ [mu' r'OIJA/mu'r�AJ it burns. /anfm'Ol'Om'Ol'O/ [an'm'Olo ID'Ol'O/ an ' mG-lGf!lG-l:rJ eood. Other varian ts of /'0/ are as in /·t'OIJkka/ [·t'OIJkAJ hat. /·t'OwuIJa/ [·t'O�WUIJAJ name of mo iety. 7

/�u'WQkkur.iti/ [�u'WQ�kUrI�I] I wi ll write, I kQPQI [ kQbQ/ 'b�bQ] foose, /'kQtQk/ ['kQqpk/'b�qpk] name of subsection. /ana'kQyp�ippre/ [ana'gQybu�Ip8/ana'gQ�ybu�Ip8] you (P l.) /an'kQlmpukQlmpu/ [an'gQlmb ugQlmbu/an'gQ�lmbugQ�lmbu] little one. /, kQkkQkkQ/ ['kQkQkQ/'kQukQuko] lonf time, /'kQma/ ['kQIIlA/'kQ�IIlA] body, /'nQkkQrQ/ ['n6kQrQ/'nQ�kQrQ] small shark /mu' rQ/ [mu'rQ] it wi ll burn. The norm of lu/ is [u] as in Imu'rumu�a/ [mu'rumu�A] broke it, /'kulukkulu/ ['kulukulu] dOf, /'lumpuk/ ['lumb uk] type pifeon , Iku'pu/ [ku ,bu] wi ll hit it. Other var iants of lui are as in /'wukkalrel ['wouka18/'wuka18] type foanna. /'�upparl ['�opar/ '�upar] cloud , I'pama 'kutirai ['paIIlA 'ku �IrA/'paIIlA 'k��IrA] its head , I'kuyu/ ['kuyu/'k�yu] nose. la'kukkukkuwal [a'koukou kuwA/ a 'kukukuWA] wi ll cool him. I'para 'mut ira/ ['parA 'mu�IrA/'parA 'm��IrA] its rea r. stern , I� u'puyanal [!Ju'b uyanA/IJu'b�yanA] I hit myself. /rJu'mu�pul [�u'moUrJbu/�u'mu�bu] I wi ll comp lete. The seque nc e luyl may be reduced to [ i] in both stressed and unstressed syllable s as in I'puyma/ ['puYIIlA/'p iIIlA] back. I�una'kuypukka/ [�una'guybukA/�una'gibukA] show me! , I�u 'mu�puyppre/ [�u'mu�buyp8/�u'mu�b ip8] I wi ll fO for fo od .

7. STRESS Stress is phonemic and contrasts as in 'ama mother and a'IDa wi ll fet him; 'k anara spoonb ill. ku'narra that one and �una'ra he will spea r me; 'k upurrQ road. ku'punQ hit it and kupu'na they wi ll see it. 'titapata pronf 'nanappuniyu his fat her ti'numurQ smel ls it ku'pupp iti�a it foes down tinu'muru mother's sister anku'kaliya nat ive man �upura'PQy we all wi ll fO �urura'kakka I wi ll pu ll it �unapiri 'ka they wi ll carry me Stress appears to be grammatically predictab le , usually occurring on the first syllable of the word stem.

8. DISTRIBUT ION OF PHONEMES All single consonants occur initially , medially , and 8

finally . The following consonant clusters oc cur word initially:

mp ) fip ) fit ) fim ) fifi ) fiy ) nt ) �k . The follow ing consonant clusters oc cur word me dially: pp ) tt ) tp ) tt ) tk ) kp ) kt kk ) r� ) lk 1m ) l� )

mp ) fip ) fit fim ) fifi ) ny ) np ) nt ) nk nm nn ) n� ) �p ) �k ) yp ) yt ) yk ) ym ) yn ) y� ) yr ) yw ) yy ) rp ) rt rr rl rn rpp ) rkk ) rmp ) lpp ) ltt ) lkk ) lmp) mpp ) �kk ) �rr ) rtt ) rtp ) rtk ) rkk ) rlp ) rlt ) rlk ) rlm ) rnp ) rnt ) rnk ) rn� ) r�k ) ykk ) ypp ) rlpp ) rltt ) rlkk ) rnpp . The follow ing consonant clusters occur word finally : rk ) lp ) lk ) rk ) rt ) rl ) rn ) rlk Example s of all consonant clusters are given in Section 12, Practical Orthography, Vowels oc cur singly, never in clusters. /a/ is the only vowel wh ich oc curs word initially exc ept in the two words 'iya and 'reywQ wh ich both indicate question. All vowels occur in word me dial and final pos itions . Finally however there is sometime s free var iation betwe en vowe l phoneme s as in /'pukulu/ , /'pukulQ/ , /'pukula/ water.

9 THE SYLLABLE There are seven syllable types each hav ing a vowel (v) or word initial syllab ic nasal as its peak and differing by the numb er and pos ition of consonants (e). They are: e as in n.'ma you (s . ) wi ll fet V as in 'a.na father, son ve as in 'an.�re who? ev as in ma fet it! eve as in PQl fire evee as in malk subsection eveee as in kQrlk swaf

10 DISTRIBUTION OF SYLLABLES Words consist of from one to ten syllables as in ta l desire and �u.na.pi.ri.ru.ra.'kak.ka.tre.na you (d l. or pl.) 9

would have pu lled me . The C, v, and Vc syllables are limited to word init ial position as in n,'pD.ya you (s .) go, a. 'pam.pu.ra. he wa lks and an. 'mD.b we ll.

11. INTERPRETATION Vocoid glides, other than those listed under Section 6, Vocalic Var iants, are interpreted on the basis of non suspect syllable patterns as vowel plus consonants y or w as in 'iya question ma rker, a' preya he arrives, an'paykkirrre long, pDy go!, a'mD!Jpuyppre he wi ll go for good, 'rawu camp, !Jaw yes . Word medially the voiceless var iants of I p J t J k I are interpreted as geminates I pp J tt J kk I. This interpreta tion is based on the voiceless qual ity of he terorganic stop clusters, the gap in the distrib uti on· pattern which would otherwise include he terorgan ic but not homorganic clusters, and the los s of voic ing which oc curs when suffixat ion cre ates a stop cluster, for example , when the suffix [-�an] feminine oc curs with ['wamut] name of subsection or with ['kD�Dk]n ame of subsection the forms become ['wamuttan ] name of feminine subsection and ['kDtan] name of feminin� subs" ection. Word medial Ipl contrasts with Ipp/ as in 'wDrlpu dragon fly and 'wDrlppu hunt!; purupu'ta:Htta they wi ll carve them and muruppu'tariya midd le aged woman; an'kapi that one and an'tappi initiate; a'pampura he wa lks and 'mampparakku your mot-her; 'IJaypurippre we, IJupura'pDy we wiII go and 'IJuppar clo ud. Word medial It I and Itt/ contrast as in paliti food and 'palitti bag; 'p Drita upland grass and 'pDrittal rice grass; 'IJata mi ld oath, 'IJattu seed of type pa lm and a'nattre sees him; a'tayhey! (f em. ) and 'pittay padd le. Word medial /k/ and /kk/ contrast as in 'pukulu wa ter and 'pukkulu forehead; 'kalikali tapping boomerang and 'kalikka li one of ma rriagea b le kinsh iP; 'mreIJkre wife, 'tiIJka panda nus fruit and 'miIJkka sandfly; 'waykiti ma y be and an'payk kirrre long. . Word medial and final [ r� J r� J r� J r+ ] are interpreted as Ir/ plus alveolar consonants, that is as / r"t J rn J rr J rl I. This interpretat ion is based on the phonetic [r] qual ity which is heard between the vowel and the following retroflexed contoid, the occurrence of Irl as the first member of consonant clusters with all point s of ar ticulat ion except the alveolar , the ab sence of contrast word initially between retroflexion and non-retroflexion and the economy of four le ss consonants wh ich result s. 10

Irtl contrasts with ItI as in 'wartu�a dOf and 'wata younf one; 'mam' part pla te, can and an'marpat ske leton. Irnl contrast s with Inl as in 'marn�a sun and 'man�a junfle; an'kurn his subsection and an'kunkun heavy. Irfl contrasts with If I as in pur'ra��i bird and pi'ra��i rice; 'marfa ta il and 'm ara comparative, Ir ll contrasts with III as in 'marla� type fish and 'kala� hook; �arl hasten ! and �al des ire,

12. PROPOSED ORTHOGRAPHY Cons onants: p � t k r 1 m n n � w y r Symb ols: p j t k d 1 m ny n ng w y r

Vowels: � re a 0 u Symb ols: i e a 0 u

Stress is unwr itten on the first syllable . Elsewhere it is symbol ised by a hyphen preced ing the stressed syllable. The cluster In�/, exc ept when divided by a hyphe n is wr itten 'nj ' rather than lnyj I because in Engl ish spelling ' y ' between consonants is syllab ic and because the cluster / nV doe s not oc cur in Burera. The following words illustrate the or thography and part icularly how it appears in all consonant clusters.

Initial Phonemic Or thographic Meaning Clus ter

mp m'pono m-pono it wen t fip fi'poya ny-poya you fO n� fi'�afi�a ny-j anja you carry it fim n'ma ny-m a you wi II fet it fifi n'firekka ny-nyekka you wi II re st ny fi'yalppu ny-y alppu you wiII cook it nt n'ta n-ta you wi ll spea r it �'kafi�a ng-kanja I carry it

Medial Clus ter

pp 'wuppu W'uppu under , tn �� 'pi��ay pi j j ay padd le 11

Medial Phonemic Or t ho�raphic Meanin� Clus ter

tp 'watpar watpar type oak tree tt tinku'patpatta jinku-patpatja type insect tk a'kutkuttiTja a-kutkutjinga he runs , trots kp 'parakparak padakpadak fro�mo uth bird kt pala'muktu pala-mukju close your eyes ! kk 'rakka rakka sit down rIJ munIJu'larIJular munngu- ladngulad li�ht one, bread lk 'kalku kalku flyin� fox 1m 'talmuru j almudu wood , lo� lIJ 'tolTjo jolngo smoke mp 'pampay pampay old woma n rip 'IJanpak ngany pak armlet fit 'kuntoIJ kunjong tree nm 'mafunak manymak ri�ht , true nn apurifi'firekka apud iny-ny ekka they 2 (fem. ) wiII res t fly pififl'yalppu pidiny-yalppu you 2 (fem. ) cook it! np an' para an-pada river mouth nt 'panta panta l e� nk 'ranku ranku moon nm a'yrenrIJH;ya a-y enmeya what tS he do i n�? nn an'nuru an-nudu ma le nIJ 'manrJa mannga jun� le IJP a'muIJpu a-mungpu wi II comp lete it IJk ,tiIJka jingka pandanus fruit yp ay'pay ay-pay you and I wi II ea t yt ay'talappuro ay-ja lappudo you and I ma ke it yk 'waykiti waykiji ma ybe ym ay'ma ay-ma you and I wiII �et yn ay'na ay-na you and I wiII see YIJ aY'IJukkurifltiIJa ay-ngukkud injinga you and I turn it ay-rudakakkije you and I pu II it yr ay'rurakakkitre yw mun'maywu mun-maywu old 12

Medial Phonemi c Orthographic Meaning Cluster yy ay'yalppu ay-yalppu you and I wiII cook it rp '�arpar jarpad Cypre ss pine rt 'wartulJa wartunga dOf rr an'lJarrappa an-ngardappa one rl kun'korlo kun-korlo dry , bare , dead wood rn 'murna murna f infe r, hand tpp 'wurppan wudppan emu rkk '�arkka jadkka water foanna rmp 'yrermpre yedmpe husband , wife's brot he r lpp 'yalppu yalppu cook it l�� m' pal��Hre m-pa ljjide hunf it up lkk pi'palkki�i pi-palkkiji it sticks to you lmp an'palmparre an-palmpade short mpp 'mamppulJku mamppungku your mothe r lJkk 'milJkka mingkka sand fly lJrr 'lJanarralJrralJ nganardangrdang Jaw r�� 'yur��i yurjj i run! rtp ku'nartpa ku-nartpa that one rtk ku'yartku�umuro ku-yartkujumudo cuts it rkk mU'lJarkkurumuro mu-ngarkkudumudo lifhts frass fire rlp 'worlpu worlpu drafon fly rl� a,kurltilJa a-kurljinga he 'Vomits rlk kuna'korlkara kuna-korlkada mud rlm a,karlmuno a-karlmuno he fot up rnp 'rarnpa darnpa thifh rnt rmrerntre mernte arm rnk 'marnki marnk i underst and inf rnlJ 'marnlJa marnnga sun rlJk 'marlJkurl marngkurl w oome ra ykk an'paykkirrre an-paykkirde lonf ypp 'lJayppre ngayppe I rlpp 'worlppu worlppu hunt ! 13

Medial Phonemic Or thographic Meaning Clus ter rlt"t anku'IJar1tta anku-ngarljja wh ite rlkk mun'porlkki"t mun-porlkkij skid ma rk rnpp 'purnppo purnppo moth

Fina l Clus ter

rk 'palmark palmadk wind Ip tolp jo l p salt Ik 'mreyrelk meyelk woman rt 'mampart mampart plate, billy can rk ' "tii'IJurk jidngurk jOf rl n'"tarl ny-jarl you wi II hasten rn mrern mern cold rlk korlk korlk swaf NOTES

1. The spelling of Burera is retained to agree with Govern ment records and already-published materials. Speakers of the language, however, pronounce the name of the ir language [bu'rarAJ, which would be spelt Pu-rada by the Pr oposed Orthography.

2. See Section 11 for interpretation of word medial voice less stop as a geminate.

JEAN F. KI RTON

O. Introduction

1. Phonemes 1. 1 Consonant De scription 1.2 Consonant Distribut ion 1.3 Vowel Description 1.4 Vowel Distribution

2. Un its Larger Than the Phoneme 2. 1 Phonological Syllable 2.2 Phonological Word

O. INTRODUCTION The purpose of this paper is to give a description of the phonology of the Anyul a1 language with particul ar reference to the phoneme, syllable and wo rd levels. 2

l. PHONEM ES Anyul a has 20 consonants which include stops, nasal s, laterals, semivowels and the vibrant r. These contrast as in Chart 1. There are 3 vowel s wh ich contrast ac cording to front, mid and back ar ticul ation. Al lophonic variation is a feature of vowel manifestation.

l. 1 Consonan tal Descrip tion

Chart 1

L abi al Inter- Al veo- Al veol ar Ret ro- Ve l ar den tal den tal f1 exed

Stops b g dj d d 9 Nas al s m n nj n n IJ Latera l s 1 lj 1 1 Semivowel s /Vi bran t w y r r

Kirton, J.F. "Anyula Phonology". In Glasgow, D., Glasgow, K., Kirton, J., Oates, W., Sommer, S.A. and Sommer, E.G. editors, Papers in Australian Linguistics No. 1. A-10:15-28. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-A10.15 ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. 16

1. 11 Consonantal Contrast Stops contrast as in: dubaga 'grinding stone' banj dja 'red ant '

wugayi 'grandchild' - 'spear shaft ' dambira. gudjaga 'mother ' djandini 'that ( fern. ) ' gudargu 'brolga' dangudangur 'red'

gu andu -- � ' hard ' dumandalu. 'to get ' wugawu 'word ( genitive ) ' gambaa 'in the sun '

Nasals contrast as in: ::umanIJu 'single man ' namba 'there ' wunaga 'younger sibling ' IJanga 'where? ' wunjaga 'father ' banj dja 'red ant ' wunala 'kangaroo ' manga 'body ' awuna , fog' ma�ga 'grub ' maIJara , food' gaIJga 'because'

Laterals contrast as in: nalanu 'ti-tree bark ' walja 'dugong ( sea mammal ) , yala 'creek ' abaral 'black cockatoo ' 'new foliage ' mawabal 'wild potato '

Laterals IN and I �I contrast as in: LJali 'what?' walja 'dugong ( sea mammal ) , IJali 'we two ( incl. ) , walgur 'asleep ' ya�a 'creek ' a�gu , fi sh' v IJaH 'really! ' argu 'other ' aga�i ' at a we stern place' -muruma 'to shut ' -muruma 'to cut '

Semivowels contrast as in: 17

walgura 'big' niwanj i 'animal fl esh' yalguyi 'young fellow' miyadji 'widower ' �awurgi ' water seep age ' wiraIJu 'concerning'

Labials contrast as in: baraga 'mother' malala 'emu ( bird) ' waraba 'no'

Interdental s contrast as in: IJana1 'body hair' nanawawidj i 'married man ' nalanu 'ti-tree bark ' lambidji 'wind'

, IJagara 'we two ( excl. ) -dambira. ' spear shaft '

Al veodental s contrast as in: dj ilili 'spring of water ' bulawidji 'white hair' njiba 'calm' -manganinja past continuous tense 'rosella ( bird) ' yiliH 'blood' ramiyimiyi 'a fl y'

Alveolars contrast as in: dangudangur 'red' langalanga 'qui ckly' nangawa 'lagoon' raIJadj ara 'lightning'

Retro fl exes. contrast as in: 3 IJ�a 'I' ya�a 'new fo liage ' mada , al so ' awara 'place'

Velars contrast as in: 'because' 'a cold' 18

1.12 Con sonan tal Var iants The series of stop s vary according to vOIcIng, voiceless ness and voi cele ss asp iration, thi s latter feature being more ap parent in sp eakers who have had most contact wi th English. Th ese allophones occur in free variation but the tendency has been noted for voicing to be more frequent between vowels and at the on set of a stressed syllable in word medial position. /dj/ is manifested by two variants according to release. The unreleased variant occurs in syllable final position and the released variant elsewhere. (unreleased) wadjbaya 'call out! ' budidjbudidj 'tangled' (released) djilili 'spring of water' badjiwa 'there he is' There is variation of a restricted type betwe en the phonemes /dj/ and /g/. The restriction is morphologically predictable. Variation occur s in ini tial position in pronominal verb prefi xes. In an area of noun affixation gi and dji- occur, the former being used exclusively by male speakers and the latter by female speakers. It may be that the fluctuation of phonemes / g/ and / dj/ has resulted from a breakdown of a similar previous distinction in usage of verb prefi xes. giyawiIJga / djiyawilJga 'he walked' galuwadjaIJga / djaluwadjaIJga 'they fished' /9/ is mani fested by [d] in word initial position. I ['njadu'ma YA] /nja�umayA/ 'g�t him! '

['dumAn'd_. -alu] /dumandalu/ 'to get' .-- ['a�i'yaIJo] /a�iyaIJu/ 'new (fern. )' [ti'yaIJo] /�iyaIJu/ 'new (masc. male sp eaker)' In intervocalic position /�/ has the free variants [d] and [f]. ['a�u]/ ['afu] /a�u/ 'girl' [tjA'ga�]/ [tjA'ga.fN /djaga.�a/ 'a group ' /nj/ has unreleased and released variants wh ich occur un released in syllable final position and released elsewhere. (unreleased) ganjma�a 'two' ( released) njigunja 'your (sing. rnasc. subj .)' 19

Preceding the velar stop, /nj/ is manifested by a backed variant [njJ or less frequently by the alternate varian t [�J. A single instance of this variat ion has been recorded in wo rd final posit ion. ['k�aInj' gahJ / [';kana.I�' gahJ / g�anjgaH/ 'I heard' [ ,tj aIn.jkAJ (,/ tj aI�kAJ /dj anj gal ' stone' ['kiyinj'giyinjJ/['kiyi�'giyi�J /giyinjgiyinj/ 'fly ing fox'

The nasals /n/ and /nj/ have a voiceless allophone in wo rd initial consonant position in the syllable type CCV.4 [Nta'�oruJ I nda�urul 'your no se ' [NpaI'yarAJ /nbayar�1 'was falling' [NJkArinj'djarAJ /njgarinjdjaral 'was hearing' /r; has the variants [rJ, [FJ, [�J, and LaJ. [rJ is the usual manifestation of this phoneme in carefully arti culated speech. [fJ is a free vari ant in emphasised utterances and in wo rd final position. [�J occurs as a free variant in word initial posit ion, and in carelessly articulated speech it varies freely in all positions other than in a consonant cluster. For example, if the sp eaker is weary [rJ tends to be the mani festation of /r;. LRJ may be the �ar i ant in phrase final position or in cluster wi th a voiceless stop . [, y uru' djuruJ / yurudj uru/ 'a long way' [ , �IlIlA'�aItj i J / [ 'h�' �aItjiJ /H ma�adj i/ ' tho se two' [, w�Injtj irJ / [ 'waI�jtj iF ] / [, waInjtj i R] / wanjdji r/ 'leaf' [:A'wargIJ/[�A'waRkI] /�awargi/ 'water seep age '

1. 2 Consonant Di stribu tion Consonantal occurrence may be described ac cordi ng to di s tribution into the phonological word. Wo rd initial: Singly, all consonants except /lj/, /n/, and /1/ occur . Recorded di-clusters are restricted t; /nb/, /�dI, and /njg/, the nasal always being voiceless. Wo rd medi al: All consonants are found in intervocalic posi tion. The following di-clusters occur: ( a) two stops: only /djb/, /djg/. ( b ) nasal and stop : Th ese may be homorganic at all six points of articulation as shown in Chart 1. In 20

heterorganic clusters Inj/, In/, /�/, and I� precede Ibl and I g/. I� and /� precede Idj/. ( c ) two nasal s: Inj/, In/, I�/, and IIJI precede 1m!. Inl and In! precede IrJl. IIJnjl al so occurs. ( d ) 11-1 and Ir-I ini tial clusters: III and IN precede I b/, I g/, 1m!, I rJl, and Iw i. Addi tionally I rnjl has been no ted. ( e ) retroflex- initial cluster s: 1:1 precedes the nasal s 1m!, and / nj/. I +I precedes I g/.

Ch art 2

CONSONANT DJ-CLUSTERS

b dj d d 9 m nj IJ w dj djb djg m mb

n ml nj njb njdj njg njm n nb nd ng nm n nb ndj nd �g nm � IJdj IJg IJm IJnj 1 lb 19 1m lIJ lw 1 �g v rb rg fro rnj rw r rm rnj

Ex ception: Three consonant clusters. Only two wo rds have been consistently phonetically recorded as ha ving a 3- consonant cluster: [A'bIr�go'bIr�go] l abir�gubir�gu l 'kingfi sher ' [mA'wur�ku] Imawur�gul 'fi shing line' Because of the strong pressure for stress to occur on the penul timate syllable, the suspect sequence Ir�gl is analysed as being a cluster of 3 consonants. Wo rd finaL: Consonants ar e less frequent than vowels in this position. IN and III occur jTlo st commo nly. Idj/, Inj/, 21

I nI, and III may a1 so be fo und. Consonant clusters do no t occur.

Examples of Si n�l e Con son an t Occu rrence

In itial and medial: See 1. 11 Final: wa�ar 'mud', wudjul 'grass', bu:ridjbu:ridj 'willy- wag tail (bird) ', warinj 'egg', �urbun 'scrub', mawabal 'wild potato ' .

Exampl es of Con sonant Cl uster Di stri buti on

In itial: nbayara 'was falling', nda�uru 'your nos e', njgarinjdjara 'was hearing '. Medial: wadjbaya 'call out! ', awidj gu 'species of lizard', namb a 'there ', �anga 'where?', ginj buga 'flying fox ', wunj djur 'fine grass', dj umanj gara. 'long', ganj ma�a 'two', lanba 'nest', anda 'she', anga 'up', anmaya 'stay there! ' , �uman�u 'singl e man', awa�bul 'species of frog ', ndami�- mi�dj al 'your eyebrows ', ab i��awa�a 'bag', m�ga 'species of grub ', wu��ar 'smoke ', nanulawuri�dja�u 'river cro ssing', baru�gu 'earl ier today', wi�i�mar 'species of gum tree', miri�nj u�u 'a lying position ', mayulbu 'rope ', walgur 'asl eep', aralmur 'kookaburra' , mimaluu 'dark of night', djulwiriri 'hawk', a�gu 'fish', :rirba:rirba 'seagull ', argula 'one', wurmul 'lazy', nandayirnj i 'her fingern ail ', yur�u manga 'continuously', am irwa 'crocodile', wirmur 'spear prong', abi�nj i 'wax'.

1.3 Vowel Description The three vowel phonemes are Iii, I ai, and lui. The back vowel is wr itten phonemically throughout as lui for convenience. Chart 3

Fron t Mid Back

i a u

1.31 Vo wel Contrast The three vowel s contrast as in: 22

giwiIJga 'it carne (non- food- class subject)' gawiIJga 'you carne (sing. )' guwiIJga 'it carne (food- class subject)' IJali 'we two (incl.)' IJala 'but ' IJalu 'south'

1.32 Vocalic Variants

Chart 4

Fron t Cen tral Back

Hi�h

Low Hi �h

Mi d

Low Mid

Low Low

Each of the vowel phoneme s has several al lophones as illustrated in Chart 4. There is free va riation of the non glide allophones within each phoneme but certain tendencies of occurrence have been noted. Additionally, each vowel phoneme has a lengthened allophone condit ioned by occurrence in a stressed syllable. Iii has the variants [i], [I], and [eL h] tends to occur contiguous to Idj/, Inj/, Ilj/, and Iy/, and in stressed syllabl es. [e] tends to occur contiguous to re troflexed consonants and [I] el sewh ere. Ial has the variants [a] , [A] , [g] , and [aI] . [a] tends to occur in stressed syllables except preceding alveodental consonants. [a�] precedes Idj/, Inj/, Ilj/, and Iy/. [A] tends to be found in unstre ssed syllables an d [g] in less carefully ar ticul ated speech. lui has the variants [u] , [0] , and [OI). [OI] precedes Idj/, Inj/, Ilj/, and Iy/. [0] tends to occur contiguous to a vel ar consonant. El sewhere there is free variation be tween [0] and [u] . Vowel allophones are illustrated below: ['maInjtji'garA] Imanjdjigara/ 'brother's wi fe (woman sp eaker ) , [:e'dinjtjA] I:idinjdja/ 'yesterday' 23

['kilI'lidjA] I gililidj a( 'he sent it' ['WoItjPI] lwudjbil 'egg ' [rA'bulbul' gidjA] Irabul bulgidjal 'species of bird' [ ':,ogo] I:,u gu/ ' dry'

1.4 Vowel Distribution Distribution of vowels into syllables has the one re stri ction th at I aI is the only vowel found in syl lable typ es V, VC. Wi thin the wo rd, Ia! only occurs in wo rd initial posi tion. Al l vowel s may precede or follow al l consonants wi th the two exceptions that Iii has not been recorded fo llowing I n/ nor I ul fo llowing / Ij/. Sequence laa!: Th is is regarded as two adjacent vowel s rather than a singl e long vowel because (i) in slow speech there is a clear re-initiation of the vowel, particul arly where the second l al is in a stressed syllable; (ii) wh ere the morpheme '-§, is su ffixed to a stem final lal, the affix -� ac ts as an additional syllable to cause forward shift of penultimate stress. For examp le, 'yinda 'you (sing. )' plus -� (accomp animent marker) becomes yin'daa 'with you'. In faster sp eech, parti cularly wh ere the second Ia! is in an unstressed syllable, it tends to occur as length on the previous syllable or may be lost.

2. UNITS LARGER THAN THE PHONEME A full study of the Anyula phrase or breath group has not yet been made. For the purpose of this paper the phonologi cal phrase may be defined as a wo rd or group of wo rds occur ring between pauses and carrying an intonat ion contour. Th e remainder of this section will be given to a description of the phonoloe;ical syllable and phonological word in Anyul a.

2. 1 Phonological Syllable A phonological syllable may be defined as having a nucleus consisting of a single vowel and potential for the occurrence of prenuclear margin consisting of one consonant or a cluster of two consonants, and a single consonant post nuclear margin. Syllable margins may be perceived by di s tribution of segmental phonemes, that is, Ill , IV, IfI, Im/ , In/ , Inl , and I ry.1, may occur as consonan t of a VC syllable; nasal s, Idj/ , Ill , IV , Ir/ , an d 1r;1 occur as 24 postnuclear margin in a eve syllable. ( For vowel s see 1.4.) Postnuclear syllable borders may be perceived by a poten tial change of intonation level and potential stress occur rence.

SyLl able Types and Di stributio n The mo st frequent forms are ev and eve, and these, with V an d eev, may occur in any position in the phonological word. ve syll able s are found only in word ini ti al and medial positions. The vowel of V and ve syllables is always /a/. V syllable: a.lu 'they ( plural ) ' dj i.ya.a.ga.H 'he was cold ( woman sp eaker) ' wun.da.a 'in the tree' ve syllable: an.ga 'sea' ga.al.gal.banj.dji 'you are washing ( clothes) ' ev syllable: bu.dj i.ma.la 'rainbow' eve syllable: wal. gur 'asleep ' na.wim.bi 'bee ' eev syll abl e: nda.wi.ni 'your name ( sing. ) ' a.bir.�gu.bir.�gu 'kingfisher '

Syllables may vary non- contrastively in length accordi ng to their position in the phonological wo rd and phrase. Greatest length is found in the nucleus of the phonological phrase and shorter length in non-nuclear position. There may be a voiceless al losyllable in syllable final position in the phonological phrase.

2.2 Phonological Word A phonological wo rd in Anyula is defined as a minimal utterance, wh ich, ac cordi ng to current dat a, consists minimally of two syllables and maximally of ten syllables. 5 25

It may al so be defined according to certain border phenom ena; that is, syllabl e di stribution, potential lenisness or devoicing of the final syllable and potential variant phon emic and allophonic manifestation in word initial position. (For detail see 1. 2, 1. 4, 2. 1. ) Stres s: Stress is heard as sl ight intensity associated with length. In wo rds of four or more syllables wh ere more than one stress occurs, there is no clear regular di stinction between primary and secondary stress. Differing intensity of stress is therefore taken to be non- contrastive. The following rul es of stress have been formul ated on the basi s of predictable stre ss behaviour and on the basis of trends noted: (1) Stress never occurs on the final syllable nor do two stresses ever occur on two successive syllables. (2) In 2-syllabl e words stress is initial. , adu 'girl' '�awu 'cloud' '�urbun 'scrub ' (3) In 3-syllable wo rds stre ss is penultimate, wi th a tend ency noted in some wo rds fo r stress to vary from penulti mate to ini ti al . ni ' wanji 'animal fl esh' IJam'bal a 'we (plural incl. )' ganj 'mada / 'ganjma�a 'two ' (4) In 4-syllable words stress occurs initially and on the penultimate syllable. 'maru' wara 'cousin' ' al anj ' dj ilu 'to the camp ' ' yur�u 'manga 'continuously' (5) In words of five or more syllables stress occurs: (i) on the penul timate syllable; (ii) word stem initial, unl ess this immediately pre cedes the penultimate syllable; (iii) word initial, excep t wh ere there is a singl e syllable prefix. (Underl ining below indicates prefixation. ) 'maruwa'ral a 'with the cousin' �'yilwi ' yilwi 'duck ' 26

�ji'walanj 'mandji 'it is emerging' 'H��'ram��a'ninj a 'he was hitting himsel f' 'H��!�'wuturu'manj dji 'we are �l eating' A tendency has been noted for stress to occur on al ter nate syllables preceding the penultimate syllable, and there may be non- signi ficant variation from the ab ove rul e ( ii) to thi s. The word for 'he was hitting himsel f' in the above examples may be gtressed as follows: 'H��a'm��a'ninjdj a. It will be noted throughout that the strong constant stress pressure is for the occurrence of penultimate stress. Where the above rules of stress ar e kep t, stress is con sidered to be predictable and therefore does not require to be indicated. Wh ere there are exceptions an d stress is therefore unpredictable stress wi ll be wr itten and indicated by the symbol 1·1 preceding the stressed syllable. NOT ES

1. Anyul a is the language of an Australian Aboriginal tribe of about 150 people who are mainly situated at or around Borroloola in the Gul f Country of the Northern Territory. The Aboriginal settlement at Borroloola includes Aborigines from a numb er of other tribes, but mo st notably from the Garawa, who, in the we t season, may outnumber the Anyul a. Certain Gar awa words are in such common use that young adul ts consider them to b� Anyula e. g. wabuda 'water'. A few Kutanj i words ar e al so known to be in use. A form of English is spoken by the maj ority of the Anyula but their own language is virile. The extent of influence of these other languages on Anyul a is not known. In Anyula there is some affixal variation determined by the sex of the speaker. Language examples in this paper are from both men's and women 's sp eech. The an alysis of thi s paper wa s done on data obtained during 12 months' field wo rk at Borroloola under the auspices of the Summer Institute of Lingui stics. Concen trated an aly sis was done at the S. I.L. Work shop at the University of Queensland, attendance at wh ich was facili tated by the Au stralian Institute of Ab original Studies. The author would al so like to acknowl edge gratefully the help of her partner Mrs E. MacDonal d, her informants Bella and Mavis Timothy, and the S. I.L. lingui stic consul tants.

2. The concept of language as a hierarchical structure which may be an alysed at different level s is taken from Kenneth L. Pike (1954 , 1955, 1959).

3. Alternatively the retro fl ex consonants I d/ , Inl , and I II ' may be analysed as �C wh ere I�I is the unit of r�troflexi�n and C the al veolar consonant. Thi s gives greater economy of phonemes. However, analysi s of the series as unit phoneme s has been preferred for the following reasons: (i) Phonetically these are heard as singl e units. I � and Irnjl are phonetically longer than Id. /, I�, and / � . (ii) Phonemically they pattern as do the single conso nant units. Th ey are found in phoneme positions

27 28

comp ar able wi th the corresponding series of inter dental , alveodental and al veolar consonants. In word final position there is only basis fo r occur rence of a single consonant and III is found in thi s position.

4. [ dn] has been recorded as a vari ant of I nl in a few in stances from one or two of the old men who have less fre qu ently helped as informants. e. g. [ , la

5. The one known exception to this is ['ye:] Iyil 'yes' wh ich seems to be of Engl ish origin. But even this occur s as Iyiyil in the usage of some of the older men.

BIBLIOGRAPHY

Capell, A. n. d. "Languages of Arnhem Land", Oc ean i a vol. 12, pp. 364- 392.

Pike, Kenneth L. 1954 Language in Rel ation to a Un ified Th eo ry of Hum an Beh avi or. Glendale, Ca lif. ; Part I 1955 Part II 1959 Part III.

W. J. OATES

O. INTRODUCTION Some light is thrown on problems pertaining to the phonemic interpretat ion of phonetically complex consonants or consonant sequences in some Au stral ian languages by a discussion of a kin problem in a New Gu inea Highland lan guage. Darlene Bee in a paper enti tled "Usarufa Distinctive Features and Phonemes"! offers two solutions to a problem in the phonemics of Usarufa concerning the interpretat ion of the sequence glottal plus consonant and of long nasal s. 2 Solut ion I ( Cluster Solution) reads: " ... interpret sequence of glottal plus consonant as clusters of two diverse phon emes". Solution 2 ( Unit Solution ) reads: " ... interpret sequence of glottal plus consonant ... as single unit phon emes". 3 The phoneme s of the two solutions are as fo llows:

SOLUTION 1 (Uni t So lution) SOLUTION 2 (Cluster So lution) p t k ? P t k ? ?p ?t ?k m n m n ?m ?n

W y W Y ?w ?y

In summar ising the merits of each solution Bee says that if Solution 1 is fo llowed it gives a simple complimentat ion statement fo r seven consonant phonemes and a singl e glide phoneme (j?/). If Solution 2 is followed it gives an in crease in the numb er of phonemes but "the increase is balanced by an exceedingly simple statement of syllable structure wi th no problem as to borders. Al so, some aspects of morphopho nemic change are mo re easily stated with this

Oates, W.J. "Syllable Patterning and Phonetically Complex Consonants in Some Australian Languages". In Glasgow, D., Glasgow, K., Kirton, J., Oates, W., Sommer, S.A. and Sommer, E.G. editors, Papers in Australian Linguistics No. 1. A-10:29-52. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-A10.29 ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. 30

interpretat ion". 4 Bee leaves to the objective evaluat ion of the reader wh ich of the two solutions is to be preferred. Attention is drawn to Bee's classification of Usarufa phonemes into four classes: Consonants, Vowels, Liquids, and Gl ides. Consonants described as consonantal plus vocalic minus fill a consonant slot; Vowels described as vocal ic plus consonantal minus fill a vowel slot; Liquids, both consonantal and vocalic plus may fill either a consonant or a vowel slot; but Gl ides, being both consonantal and vocalic minus fill neither a consonant nor a vowel slot. In other words, they are undefined as to C or V class excep t to say that they are neither. Because it has been left to the reader 's choice as to whether glottal plus consonant is a single complex phoneme or a cluster of two diverse phoneme s, here too there exists an ar ea wh ich is unde fined by CV analysis, not as to class but as to wh ether there are one or two uni ts. Bee's paper is mo st valuable in so succinctly expo sing this problem of the interpretation of undefi ned data. It is not always possible to set up a grid from non suspect C and V patterns into wh ich al l data can be fi tted on that level of analysis without either forcing it or coming into conflict with other phonemic premises. If, when deal ing with unde fined data, mo re than one level of analysis is allowed fo r, the confl ict between two phonemic procedures (that of distribution of phones and phonemes and that of interpretat ion) is el iminated. If two levels of analysis are accepted fo r Bee's material, both her So lutions 1 and 2 can be adop ted together thus giving the benefi t of simplic i ty of each solution at their respective level s: Solution 1 on the segmental level, Solution 2 on the syllable level . The procedure is visualised as set out in the fo llowing chart:

LINGUI STIC LEVEL DATA PROCEDURES CONCLUSION

Syllable Etic Syllable Distribution wi th Ernic in CV syllable Syllables i l' CV grid CV analysi s (pos- sibili ty of some slots undefined) l' -.v Segmental Phonetic Data Separating and uniting procedures Phonemes 31

In app lying this procedure to an Usarufa word like k6?6?ke?A every kind, a problem is reached wh en the sequence ?k occurs, as no non- suspect pattern in the language gives clear indi cation as to wh ether thi s is CC or C or neither; it is an undefined slot. If glottal existed in the language only in such sequences as ?k, the interpretation of such sequences wo uld be clear, but conflict of interpretation arises because of the occurrence of glottal el sewhere in the lan guage in mutually exclusive di stribution with the manifesta tion in a sequence. Diagrammatically the problem and its solution may be represented as fo llows:

LINGUISTIC LEVEL INTERPRETA TION CONCLUSION

Syllable Of syllables CV CV CV CV Syllable units cv grid ( estab- cv CV (?)V CV One un- lished on non- defined suspect data) slot

Segmental Of segments CV cv ccv CV Phonemes k6 ?6 ?ke ?).. every kind

This solution suggest s that fo r ease of descriptipn the undefined slot be considered as two phoneme uni ts on the segmental level and one syllable unit on the syllable level . 5 It will be noted that the CV grid is set up en tirely on non- suspect dat a. If it can be shown that the questionable slot in the grid is not really questionable, doubt of its interpretation is removed on both the syllable and the segmental level . But wh ile doubt exists an arb i trary either/or decision ( one or two ) results wh ich at this point in the an alysis is neither desirable nor necessary. Following the above an alysis the Usarufa phoneme chart would be as fo llows: 6 SEGMEN TAL PHONEMES SYLLABLE ITEM S P t k ? ?p ?t ?k m n ?m ?n w y ?w ?y

We will now proceed to show how such a principle assists 32 in describing some areas of phonemic analysis in Austral ian languages.

1. PREDOMINANT SYLLABLE PATTERNS IN AUSTRALIAN LANGUAGES There are two prevailing non- suspect syllable patterns commo n throughout Au stral ian languages, ev and eve. In some languages as Gugu-Yal anj i (Cap e York) comb inat ions of these two patterns comprise 90� or more of the wo rds of the lan guage. They are the only two patterns noted in other languages such as Yugambe7 and Walbri8 . In Chart A examples are listed from fi fteen representat ive languages in wh ich these syllable patterns occur.

2. SYLLABLE CONSONANT/cONTOID FILLERS In many Austral ian languages the consonant fillers of these two prevailing patterns manifest certain common characteristics. The characteristics pertinent to the thesis of th is paper are here listed. They divide the fillers of the consonant slots into four different con so nant types: (i) The almo st universal phenomena of medial clusters of two emic (non- su spect) consonants9 (e type 1- 2) . (ii) But note, some languages manifest etic clusters of three and some of four medial contoids. (iii) The mo st common two conto id clu sters are stop-nasal or nasal-stop clusters with homorganic clusters (e type l-2A) occurring statistically more frequently than heteroganic clusters10 (e type l- 2B). (iv) The ab sence of retro f lexed oral resonant (usually symboli sed / r/ or /r/) wo rd initial 11 with the ex ception in s �me languages of just a few words1 2 (e type 3A) . (v) Th e rare occurrence of any retro f lexed conto id wo rd initial 1 3 (e type 3B) . Their di stribution is similar to that of the retro flexed oral resonant being found usually between vowel s or in medial consonant clu sters. (vi) But note, if in a language retro fle xed oral reso nant may be found wo rd initial, then usual ly the other retro fle xed contoids al so occur wo rd initial, but if retroflexed oral resonant does not occur 33

word initially, so usually neither do other re troflexed contoids. (vii) The frequent occurrence of retro fl exed oral reso nant syllable finally1 4, usually wo rd finally1 5 (e type 4).

3. INTERPRETATION OF HOMORGANI C NASAL-STOP SEQUENCES

3. 1. Problems of Interpretation in some la nguages (such as Gugu-Yalanj i) There is plenty of evidence to show that homorganic nasal stop sequences are very close-knit in Austral ian languages. In some languages the que stion ar ises wh ether they function as one unit as in Ar anda and as M. C. Cunni ngham has inter preted them in Al awa 16, or as a sequence of two units. The problem of interpretat ion arises in languages in wh ich they are part of a sequence of more than two consonants partic ularly if such sequences occur in only a smal l percentage of words as in Wik-Munkan and Gugu- Yalanj i. In the latter lan guage in about five per cent of wo rds homorgamc nasal- stop clusters occur in a medi al cluster of three conto ids the first member of wh ich is 1) r) r) or y. EXAMPL ES: wa-lmQ-a Log; yi-rmQ-a three prong sp ea r; wa.-�!.l.9-u sLeep ; wa-rng-i1 wa ke up ; wa-��-il soft ; bu-�!l.9-UY snore. The que stion ar ises whether, to fit the permi ssible non� suspect sequence of two medial consonants, the sequence pat terns as one phoneme or wh ether it patterns as a sequence of two phoneme s. Our ini tial int erpretation 17 of regarding these em ically as two separ ate phonemes in a medial cluster of three consonants fo rming a third, limited, consonant vowel pattern, eev (and al so eeve) , was somewhat arb itrary though it proved a workable solution as trial literacy showed. (The medial nasal consonant of the cluster was taught as a continuant. ) The following analysi s suggest s a better solut ion. It postulates that: (i) the homorganic nasal-stops are syllable units mb ) nd) �g) in certain envi ronments. They are symbol i sed as NS in this paper; (ii) their structure is complex, manifesting a fusion of two component parts in some environments and so 34

function as one syllable unit, but in other envi ronments their phonetic components function as two phoneme s, ro + b, n + �, � + g, symboli sed as the /N/ and / S/ phoneme s in this paper; ( iii) the component parts of the NS syllable unit both retain their phonemic status wh en viewed solely on a segmental basis.

3. 2. In terpretation in Gugu- Valanji of the NS Syllable Unit The postulation of the NS syllable unit in Gugu-Yal anj i is based on two non- su spect patterns, ev and eve, wh ich limit the permi ssible wo rd medial consonant cluster s to two based on the non- suspect pattern of no t mo re than two . Thus in the above conto id cluste rs of three, the homorgani c nasal stops are interpreted as syllable units rob, nd, �g, filling a single consonant slot.

3. 3. Binary Na ture of the Homorgan ic Nasal-Stop Pho nemes in Gugu- Valanj i The NS units exhibit complexity in that in some env i ron ments their component parts function as separate phoneme s. On the phonetic level there are five ph?nes in the homo rgmric nasal- stop series, N, S, NS, N- ( S), (N} - S. On the syllable level three contrasts exi st, /N/, /S/ and NS. NS is in phonologically pre dictable variation wi th /N/ + /S/ wh ere IN/ is homorganic with regard to point of articulation wi th /S/. On the syllable level NS fills two different types of syllable slots, a ee consonant slot and a e consonant slot, creating two ( wi th a possible third) type of homorganic nasal-stop sequence as fo llows: ( i ) N- s, (ee) type wh ere the nasal fills one consonant slot and the stop another. It occurs excep t fo l

lowing r , r, 1, or y and the first consonant of a suffix morpheme. Phonemic Evi dence: Based on the non- suspect cluster of two as in gunQa finish. the rob as in gamQi clothes patterns as a sequence of two consonants. M orphophonemic Evi de nce: Mo rpheme breaks between nasal s and stops suggest the nasal closes one syl lable and the stop opens the next. Compare: bunday sit. bun-gan=gay sittinf; du�ay wen t. du�an-gu�ay fo inf. ( ii) NS2 (e) type where, based on the total perm i ssible sequence of consonants as two , the homorganic nasal- 35

stop sequence fills one consonant slot. It occurs only following r} r} 1 and y or is the first consonant of a suffix mo rpheme. Phon emic Evi dence: wal-mba log, war-�2u sleep, bur-�guy snore, way-mbil soft . Morphophon emic Evi den c e : In the fo llowing selection of locative allomorphs, mb and �d appear to fill one consonant slot following the cv shape of the majority: -ba} -bu} -�a} -�u} -mba} -nda} -ndu locat ive. Comp are buwun-ba into the boat, gabay-mba on the ant bed (both CV-CVC- CV sh ape) ; bana-IJa for water, bambu-ndu to Bambu (CVC (C) V-CV shape) . (iii) N-S3 or NS3 is,po stulated as a third type to ac count fo r al l the data. Th is type is indefinite. It is uncl ear wh ether it is a singl e of a double consonant because the slot is not clearly defined as to wh ether it is a single slot or a sequence of two. It occurs other than in the two envi ronments listed above. In a wo rd like wa�guriga to ask , there is doubt wh ether the syllables are wa-�gu-ri ga or wa�-gu-ri-ga. Thus NS consists on the syllable level of three al lo types, N-S1 and NS 2 wh ich are in mutually exclusive distri but i on, an d N- S3 or NS 3 wh ich is regarded to be in free variat ion with N-S1 and NS 2 in a third environment.

3 . 4 . Phonem ic Status of Component Parts of the NS Unit in Gugu · Yaianj i The component parts of the NS unit both retain their phonemic status even when they occur as NS 2, each component being an allophone of the phoneme in isolat ion. Where N 1 represents the nasal of NS and N2 represents other nasal s, N1 occurs only preceding a homorgamc stop, N2 occurs else where. Where S1 represents the stop of NS and S2 represents other stops, S1 occurs only fo llowing a homorganic nasal, S2 occurs el sewhere. Compare: gambi clothes, gami grandfather, gaba ra in; ja�9.a large stone, mi�u hook, jagal overflow. Structurally NS is similar on a morpho logical level to morpheme s wh ich may be bound or free. For example, the comp onent parts of I wi ll fall into separate morphological slots (cp. N- S1) , but in I'll they ar e fu sed into the one morphological slot (cp. NS 2 ) , though the component I still 36

retains its pronominal status and the component 'll retains its verbal status. In summary, in Gugu-Yal anj i there are consonant s wh ich operate on two level s: (i) on the phoneme level nasals and stops func tion as the phonemes /N/ and /5/; (ii) on the syllable level the homorganic nasal-stop sequence may manifest itsel f as a combination of the phonemes /N/ + /5/ or as the syllable unit, NS.

4. INTERPRETATION OF RETROFLEXED CONTOIDS of the no ted features of Au stral ian languages is the One 18 series of retro fl exed conto ids. Traditionally these have been considered as phonetically simple conto ids filling a single consonant slot. This analysis has been supported by the fact that it puts them into a neat series both laterally and laminally, and by the fact that as filler s of a single consonant slot they fit the customary CV patterns of the language. However, in some Au stralian languages retro flexion ex hibits similar complex characteristics to the homorganic nasal-stop series in Gugu-Yal anji in that it has closely related phonetic features of (a) retro flexion, (b) laminal feature, wh ich often manifest themselves as syllable units but may al so manifest themselves as separat e components. 19 4.1. Binary Nature of Retro flexion in Tiwi In Tiwi the phenomena of retro flexion exhibits both the above phonetic characteri stics and like the homorgani c nasal stop s in Gugu-Yal anj i fills, according to different environ ments, both a one consonant slot and two consonant slots on the syllable level . However, in Tiwi non- suspect clusters of three consonants occur wh ich allow all of the retro flexed series to fit into two consonant slots. In other language s such as Nyangumarda and Wailbri (see Chart E) it is neces sary to postulate the retroflexed series as syllable units since these languages have no larger clusters than two non suspect consonants. In the following argument, Rl represents retroflexed oral resonant, RL represents retro fl exed contoids. RL con sists of the fo llowing phonetic features: R - feature of retro flexion; L - laminal fe ature wh ich has three different manifestations: Ls, a feature of complete stricture; Ln , a 37

feature of nasal release; Ll, a feature of lateral release. On the phonetic level there are five phones in the retro flexed contoid series: R, L, RL, R(L}, (R}L. On the level of contrast and mutually exclusive distribution these form three phoneme s: /R/, /L/, IRL/20 as fo llows: IRI phon em e: The feature of retroflexion R is a sub-memb er of the retro fl exed oral resonant R1, R occurring only in association wi th L (LS, Ln, Ll) , Rl occurring elsewhere. EXAMPLES: wu!:�a they; a!:napa wait; parlini old; (cp. �ipa rieht side; ku!:iwa mo rn ine ) .

/LI phon em e: Each of the three elements of the laminal fea ture L (Ls, Ln, Ll), is a sub-memb er of each of the fOllowing phoneme s, It/, In/, /1/, respectively, occurring in mutual ly exclusive distribution with them as fo llows: Ls, Ln, Ll occur only in association with R, It/, In/, III occur else where. EXAMPLES: wur�a they (cp. ya�i one item, masculine); ta:birni yesterday (cp. mi�ani now) ; parlini old (cp. tulguli bitter) .

/ RL/ phon eme: Th is phoneme is postul ated as follows: IRLI contrast s wi th I t/, I n/ and 11/. EXAMPLES: yar.ti earth, ya�i one item, mascu line; tabirni yesterday, bamagabamini wide; pad.ini old, ta:liki toneue , kalitaH ear. /RL/ al so contrasts with /R1/. Because no contrast exists between RL and R1L, RL may be regarded as R I L since the consonant Rl is a feature of the language. In practical application the postul ated phoneme s perti nent to the retroflexed series in Tiwi are: /t/, /n/, / 1/, /r/, /rt/, /rnl , /rl/ . These are similar to the phonemes pertinent to the prenasal ised stops in Gugu-Yal anji. They ar e sim ilarly structured and their functional load is similar. 2 1

5 . OBJECTIONS TO ANALYSIS The main objections to regarding retroflexion as a com plex phoneme are: (i) it do es not fit the CV patterns of the language as well as the traditional an alysis of regarding it as 38

a phonemically simple phoneme; ( ii) /r/ being a weak semi-vowel is no t likely to in fluence stronger phonetic features such as stop and nasal contoids.

6. DEFENCE OF ANALY SIS

6.1. CV Patterning It has already been shown that medial cluster s of two consonants is a common feature of Austral ian languages and that the mo st usual place for the occurrence of retro flexion is intervocal ic. Thus the interpretation of a medial re troflexed consonant as rt, rn, r1, fits the pattern of the non- suspect consonant sequence of two established in other wo rds in the language. In Tiwi , [paliniJ 0 ld interpreted as /par1ini/ fits the CVC- CV-CV pat t erning of words like /tu1duli/ bitter, / girdj ini/ small. It has been stated that Tiwi has only two CV patterns, V and CV and consequently no consonant clusters. 22 But there is plenty of evidence that consonant cluster s exi st though not as extensively as in many other Au stral ian languages, as the following evidence demonstrates: ( i ) there are non- suspect sequences such as rIJ, 1 9 as in yir'�ani la�oon, tu:l'gu:1i salt. ( ii) there are reverse sequences as kIJ, �k; tr, rt as in irk'�a:ba mouth, ana��wa not; :trumura 'gini fast , ku1 uwar:ti dirt. ( iii) there are prenasal ised heterorganic stop s: am'gia and, yawrabu'r��ji smooth. ( iv) stress occurs between the nasal and stop in both homorganic and heterorganic stop-nasal clusters indicating the nasal closes one syllable and the stop opens the next: am'gia and, IJerari�'kiti bite, kin' gaIJa fo 0 t . 6. 2. Patterning of Irl in Non -retro flexed Stop and Nasal Sequences Further ev idence in Tiwi that the retr�flexed series pattern as a sequence of two phonemes is the occurrence of the / r/ phoneme in clusters wi th stops and nasal s other than alveolar. Thus the interpretat ion of the retroflexed conso nant s as rt, rn, r1 fits a gap in the patterning. Comp ar e 39

ara'ri�Ea ri�ht, udu ' war1a hunt (both patterning V- CV- CVC CV) j gi��.iini small; i�giri tja'pe:ya count; ka' ruina�!j 'gedi tabu; mb ir'!jari fi�ht. Similar features occur in Burera (Arnhem Land) wh ich has permissible sequences of two ini tially, medially and finally an d the occurrence of such sequences as rr, rp, rk, suggesting the retroflexed series al so fit thi s pattern. In an unpubl ished � ap er Gl asgow has so interpreted the retro flexed consonants. 3

6 . 3 . Morpheme Boundaries Wh ilst it is not necessary that syllable boundar ies be congruent with morphe�e boundaries it is awkward to postu late morpheme breaks through a consonant. The archaic form of the Tiwi mascul ine morpheme -ti is suffixed to stems ending either in a vowel or the phoneme /r/. When -ti is suffixed to a pronoun ending in r, the resultant phonetiC form is the common Au stral ian retro fl exed conto id [tJ . The element of retro fl exion bel ongs to the stem mo rpheme and the laminal element to the suffixial mo rpheme. Though structural pressures may suggest the retroflexed consonant is a single consonant / �/, other pressures strongly suggest that the two basic phonetic features of wh ich it is composed should not be submerged. EXAMPLES: -ti su ffixed to r -ti su ffixed to a vo wel mur-ti son mari-ti ra inbow snake miyar-ti pandanus ila-ti kn ife kuluwar-ti di rt kulu-ti axe 24 yar-ti �round ya-ti one item This parti cular feature has been no ted by other lingui sts in other languages; Dr von Brandenstein has noted its occur rence in Jindj iparndi and Ngarluma. 25

6 . 4. The Stability of the Irl Phoneme The fo llowing evidence is put fo rward to show that (i) in Au stral ian languages the linear division of alveolar is un stable, and (ii) that the /r/ phoneme exhibits many char ac teristics of stability. (i) The instability of phoneme s in the alveolar ,region is demonstrated by the fact that it is a common feature in Au stralian languages for t to be drawn to the alveopalatal position tj conti guous to i, or to be drawn to the point of articulation of a con tiguous consonant. For example, the Tiwi wo rd 40

/ tirti/ is pho netically [tj i�iJ , bad. In We stern Desert , the alveolar consonant of the locative and transitive verb suffixes changes its point of articulation according to the final consonant of the stem to wh ich it is suffixed: -ta and -tu be· come -la and -lu when suffi xed to a stem ending in a lateral , -ta and -tu wh en suffixed to a stem end ' ing in a ret;oflexed consonant, etc. 26 (ii) The stability of the /r/ phoneme is demonstrated by the fact that it frequently influences alveolar phoneme s to the retroflexed po int of art iculation as cited above. This phenomena is seen parti cularly in the case of the homorganic nasal stop sequence;

nd. . and nd are both commo'. n sequences but nd and nd .. do not occur. Retroflexion thus exhibi ts greater stability than the alveolar feature. We have al ready no ted in Tiwi how the conso nant t of the mascul ine suffix is changed to the point of ar ti culation of the /r/ phoneme fo llowing stems ending in r. In Gugu-Yal anj i the / r/ phoneme is very restricted, occurring alway s as a syllable final feature except fo r one example in 1000 wo rds. Re troflexion is not a feature of the language except fo r a few wo rds like waf�dil wake up and mUf�i to twist where it occurs in association with the / r/ phoneme in its usual syllable final positi on. Had the alveolar phonemes been the more stable ele ments as one wo uld expect, these wo rds would have been phonetically [wafndil] and [murni] .

6 . 5 . Retro flexion of Vo coids It is a common featur e in Au stral ian languages that vocoids are retro flexed before a retro fl exed contoid but not fo llowing one. Douglas draws attention to this feature in the We stern Desert language in the pronunciation of such words as [m�a] buttocks, [m�� angka] at the re ar, and [w��aJ tree, 27 as do es O'Grady in Nyangumarda in such wo rds as 28 [Y1:l�a] fish, [tY1:l��u] heap of s and, [tadad] mo on. Thi s indicates that retro flexion is more parti cularly associated with a precontoid feature and indicates that the element of retro flexion is not centred in the contoid else it would be reasonable to expect retro flexion of the vocoid fo llowing as well as preceding. The centre of retroflexion appears to lie rather in the / r/ phoneme wh ich is manifested as a pre contoid feature, 41

6 . 6 . Rarity of Wo rd Initial Retro flexion Investigators appear to have difficulty in recording initial retro flexed consonant s. The difficulty may reflect actual fluctuation in this region but it is significant that word initial contrasts are rarely if ever listed in the re tro flexed series. Al l such contrasts in languages such as Gunwinggu, Burera, We stern Desert, Wailbri, Tiwi are re corded in the wo rd medial po sit ion. Thi s fact adds we ight to the feature already discussed that the /r/ phoneme has strong preference fo r syllable final position ( though it does not occur exclusively here ) . It is worthy of note that wh en word initial retro flexed contoids occur in a language, the / r/ phoneme may al �o occur wo rd initially, strongly suggest ing a link between the two ( see Chart B) . NOTES

1. Linguistic Circle of Canberra Publication, Series A, Occasional Paper No .6, Canberra 1965. 2. The allied problem of the interpretation of long nasal s dealt with by Bee is omitted as being unnecessary to the present argument. 3. Page 43. 4. Pages 42-43.

5. This principle is stated in Pike's Phonemics (University of Michigan Press, 1947), Analytical Procedure IV-J: 147-8 where he states that a syllable may have a close-knit se quence of two phoneme s on the segmental level wh ich acts in distribution in the syllable like a single simple nuclear phoneme.

6. The term item is used to describe a feature of the lan guage wh ich may be non-phonemic such as stress and intona tion wh ich are often so described wh en they are non-phonEmic.

7. See: A Description of the Yu Aambe: Di al ect of Ban dj al an A , Margaret C. Cunningham, Summer Institute of Lingui stics (to be published) .

8. See: PhonoloAy of Wai lbri , K. C. Hansen, Summer Insti tute of Linguistics (in MS) . 9. See underlined clusters in Chart A.

10. In GU Au-Yal anj i, in a selection of 618 wo rds containing medial consonant cluster s, almo st 50� (298 wo rds) were the nasal- stop series and of these 60% were the homorganic clusters mb, nd, �g. Kundj en has reverse sequences of all the homorganic clusters: mp and pm; nt and tn, �k and k�. See Kundj en PhonoloAy: Wo rd, Sy llabl e an d Phon eme, B.A. and E.G. Sommer, Summer Inst i tute of Lingui stics (to be pub lished) . 11. The III phoneme behaves much like the I rl phoneme in this and other respects.

42 43

12. See Chart B. 13. /1/ is al so usually a syllable final phoneme . 14. See Chart A. 15. See Chart A. 16. Discussion in private correspondence. Material not yet publi shed.

17. Gugu- Yal anji Phonemes, William and Lynette Oates, Oc cas ional Papers in Abo riginal Studies No .2, Au stral ian Insti tute of Aboriginal Studies.

18. See Chart C. 19. Examples quo ted in this paper have come from the fol lowing sequences: own transcription of tapes the property of Dr Helen Wurm, Au stralian Institute of Aboriginal Studies, and two MS. papers, Ch anges in Ti wi Language, Arnold R. Pilling, Wayne State University, and Ti wi (or Wo ru9 uwi) Grammar, author unknown.

20. See Chart D.

21. See Chart E.

22. Stat ement from Ti wi (or Wo ru9 uwi) Grammar.

23. Th e Phonem�s of Bu rera, David and Kathleen Gl asgow, Pacific Linguistics, Series A - Occasional Papers, No . 10 , Papers in Au stral ian Linguistics No .1; 1967; pp. 1- 14. 24. Examples from Pill ing's MS. pp. 5- 6. 25. In a report to the Au stralian Institute of Abo rigi nal Studies.

26. Weste rn Desert Grammar, W. H. Douglas, pp.90, 92. 27. ibid, p. 7.

28. Nyan guma ! a Grammar, pp. 6, 16, 12. 44

Ch art A

PREDOMINANT ev AND CONSONANT TYPES

types of emic consonant clusters ( e type 1- 2) includ ing types of heterorganic nasal-stop clusters ( e type 1- 2B) ( ) homorganic nasal-stop clusters ( e typ e 1- 2A) * syllable final /r/ ( e type 4)

eve SYLLABLES AND ev- eve LANGUAGE ev SYLLABLES e TYPE 1- 2A COMBINATIONS

Gugu-Yal anj i ba come here jin-hal-min ran IJu;r*-ma shadow (Cap e Yo rk ) ga-ba rain bal-ga talk ba-da-ba-da at a (dambal shoe bi-lar* candl e lowe r leve l nandal bury nut mUIJ ga hairl

Wik-Munkan mi-na meat nan-Eal from then mal-.ea dance (Cape York) IJa-ka IIJa t e r ( kempa fl e sh kantan th ieh pUIJku knee)

Gidabul ja-li tree ku;r*-kun talk ya-ra-man horse (Nthn N. S. W. ) ba,y:-kal kal-ka-l i-wa li-wan chop pine

Yugambe ma-li that dul-gal di rty di-bir* p l ove r (Nthn N. S. W. ) ga-li this mu-li-mir* up (yugambe no the hill barundi tn the creek gUIJgi tn the wa ter)

Wunggada ma-ma fat her nal-ga qu i c k l y (Western ba-da-ba-da fieht (Western Desert ) nUn-IJa woman Desert) (IJampa eee ya-ral-ha red nantura mu lea erass maIJka h a i rl

Ungaringin pa-na-ru sandstone bo :r* handle ba-rJal bat (Kimberleys) na-ma-la hand mal-!Ji-ri l ieh tn ine

(Continued on page LJ5) 45

Chart A - continued from page 44

eve SYLLABL ES AND ev- eve LANGUAGE ev SYLLABLES e TYPE 1- 2A COMBINATIONS

Nyul-Nyul �e yes nem his eye pa-ler* my mouth (Kimb erleys) �ab father wa-ra-gan ea�le mo r*-gol wo rk hawk go,i-gor* quiet

Wo rora ma there ka-ba,i-Qa earth (Kimberleys) ma-ba 0 ld man ma,i-na-nim creek

Kokobera la ove r there bib fathe r wal-E i we st (Gul f) ma-�o-ra unde r mar* hand bir*-ga-da smoke arm hai r bun-men blood nal-ber* ton�ue ( p��nti wh at pa�geli hunt )

Aranda ma-ma father tal-Ea mouth (Central Au st. ) ma-la female ka-na-�a two brothers

Wailbri �a-ba wa ter (Central Au st. ) ma-lu kan�a ro o (njambu the re na.r*-ga man ga-di-di mouth pindi skin mil-Qa eye mi�giri ant bed) dja-�an-Qa opos- sum

Waramunga wo-ka wo rd gu� stick mun-gu be lly (N. T. ) gu-na-ba do� bul-Qul down win-Qin break

Anyula �a-la but nan-kar* co l d ku-tar*-ku (Gul f) ma-la-la little b ro l�a emu (galamba fo rehead na-war*-ki a balanda wh ite man soak baridjuIJgawa ter lily )

Gunwinggu ma �et on wi th it! yag-min finish gun-me�-Qe the re (Arnhem Land ) ba-le wh at IJa-gar -me I have it

Burera ma �et it pol Ii re ka-la-ma� axe (Arnhem Land ) la-ma shovel wa3!-12ar* t yp e 0 ak tree wa;r*-3!u-na spear (pampay old woman panta l e� ti�ka pandanus fruit ) (Continued on page �6) 46

Chart A - continued from page 45

eve SYLLABLES AND ev-cve LANGUAGE cv SYLLABLES C TYPE 1- 2A COMBINATIONS

Gubapwingu na what ! dud he sat down jul-!}u pe rson (Arnhem Land) m�la crowd bag it broke ba£*-gu far away d��al-.9u-�al make bi� fi re

Chart B

DISTRIBUTION OF RETROFLEXION IN SIX REPRESENTATIVE

AUSTRAL IAN LANGUAGES

RETROFL EXED / r/ & RETROFL EXED LANGUAGE /r/ WO RD INITIAL CON TOlD CON TOIDS INTER WORD INITIAL VO CAL IC OR IN ec

Tiwi None None tjiri�a little (Bathurst Is. ) t�injiti stone axe tabi�i yeste rday pali�i 0 ld

Gunwinggu Only a few ex did mo on �a�e I �o (Arnhem Land) amples in 1000 gun�urg cave wo rds: -nu�ime keep silent rowug al l -ronbu skin a kan�a- redname� nest ro o ruyi ripe

Western Desert A few examples: �ikara dry bark wa�u fi re riru-riru bul l tulku corroboree wata tree roare r ninti cl eve r mana buttocks rak�raka jaw rapa confident ma�u hills kan�aro o k�pi fat tu�ku son�

Wailbri None None wi�i boy (Central Aust. ) wana snake wa: lu fi re manilba hai r

(Continued on page if7) 47

Chart B - continued from page 46

RETROFL EXED /r/ & RETROFL EXED LANGUAGE / r/ WO RD INITIAL CON TOlD CON TOIDS INTER- WO RD INI TIAL VO CAL IC OR IN CC

Gugu-Yalanj i One example in None In just a few words 1000 wo rds: as: runudj i one ve rsed murni to twi st tn esoteric warn..dil wak e up knowl edge Burera rakka sit down �iyama sp . sheII tja�aI}a sand ranka moon fi sh IJa:rpar cyprus pine raman fluff I}orI}�o mud whelk wor�o sibling shell fi sh tjinkubu�a catfi sh lama shove l spear head

Ch art C

CHARTS OF PHONEMES OF TYPICAL APICAL - CONSONANT LANGUAGES

Bi - Al - Retro- LANGUAGE Den tal V el ar Gl ott al labi al veol ar f1 exed

Western p tj t t k Desert m ny n n IJ ly 1 1 r w y r

Gunwinggu b dj d d 9 ? m nj n n IJ lj 1 1 r w y r

Wai lbri b dj d 9 m nj n n IJ lj 1 1 w y r (Continued on page 48) -

Ch art C - continued from page 47

Bi - Al- R etro- LANGUAGE Den tal Vel ar Gl ottal 1 abi al veo 1 ar fl exed jNyangumarda p tY t t k

m nY n n iJ lY 1 1 r w y r

Ch art D

PHONEMES ASSOCIATED WITH RETROFLEXION IN TIWI

LATERAL DI VI SION ------)�

LAMINAL DI VI SION Al veol ar Re t ro E1 exed IRLI ILl St ops ItI rt------L s

Nasals In! rn------ln

Laterals I II tn r l------ll mutual ly exc lusive distri bution Re sonants r 49

Ch art E

CONSONANT SEQUENCES FUNCTIONING AS A SEQUENCE IN ONE

ENVIRONMENT AN D AS A FUSED UNIT IN FOUR LANGUAGES

SEQUENCE FUSED UN IT (Filling too C slots) (Filling on e C slo t) HOMORGANIC. NASAL STOP SEQUENCES CVC-CV CVC- CV GU GU- Y AL ANJ I gam-bi clot h e s wal-IJga to han� ( cp o gun-ba Ji ni sh) (cp . wal-IJa to op en ) bun-dan-day sittin� (cp. bunday sit

RETROFLEXED CVC- CV- CV CON SONAN T SEQUENCES tir-ti-ta little TIWI (cp. tul-tu-li bitter gir-dj i-ni small )

CVC- CV

yar-ti earth (cp . cv- cv yar-ti one item)

WAI LBRI CVC-CV

warl-ba wind (cp. yur-na trave l- l in�)

NYANGUMARDA kart-ku rtve r �um (cp. yar-ti late r on) BIBLIOGRAPHY

Berndt Catherine 1952 A Dr am a 0 f N or t h - e a s t ern A r n hem Lan d . 0 c e an i a vol. XXII, No s. 3 & 4, March and June 1952.

Berndt, C . H and R. M. 1951 An Oenpelli Monologue: Cul ture Co n tac t. Oceania vol. XXII, No ; 1, September 1951.

Capell , A . 1952 Th e Wa i lbi ri th ro ugh thei r own Ey es. Oceani a, vol. XXIII, No .2, December 1952. 1953 No tes on th e Wa ramun ga Languag e, Cen tral Au stral ia. Oceania vol. XXIII, No .4, June 1953.

Cunningham , Margaret C . n. d. Di al ec t of Bandj ala ng. A de scription of the Yugarnb e. MS.

Dougla s , W H . 1958 An In troduction to the Wes t ern Desert Language. Oceania Monograph, No .4. 1959 Illus trated To pi cal Di ctionary of the We stern Desert Language. Un ited Aborigines Language Department.

Elkin , A . P ( Ed. ) n. d. Studi es in Aus tral ian L ingui sti c s . Oceania Mono- graph No .3.

Glasgow David an d Kath leen 1967 Th e Phonemes of Burera. Pacific Linguistics, Series A - Occasional Paper s, No . 10 , Papers in Au stralian Lingui stics No .1.

Han sen . K . D. n. d. Phonology of Wai lbri . Summer Inst itute of Lin- guistics. MS.

Kirton. Jean F . 1967 Anyul a Phonology. Pacific Linguistics, Series A - Occasional Papers, No .lO , Papers in Au stral ian Linguistics No .1. 50 51

O ates, Lynette 1964 Di stribution of Phon emes and Sy l labl es in Gu gu Yalanj i. An thropological Linguistics vol .6, No .1, January 19 64. A Ten tati ve Descrip tion of the Gunwinggu Language. Oceania Monograph No . 10 .

Oates, William and Lynette, and Others 1964 Gugu- Yal anji an d Wi k-Munkan Language Studies. Oc casional Papers in Abo riginal Studies No .2, Aus tralian Institute of Ab original Studies, Canberra.

0' Grady , Geo ffrey N. 1963 Nyangum a � a Grammar. Oceania Monograph No .9.

0' Grady , Geoffrey N .,. Vo egel ein, C. F. and F. M. 1966 Languages of the Wo rld, Indo-Paci fi e Fasci cl e Si x. Anthropological Lingui stics, vol. 8, No .2, February 1966.

Pike, Kenneth L. n. d. I nte rp enetration of Phonology, Morphology, an d Syn tax. Proceedings of VIII Internat ional Congress of Lingui stic s. 1947 Phon emi cs. Ann Arbor. 1947 O n the Phonemi c Status of En glish Diph thongs. Language, vol. 23, No .2, Ap ril-June 1947. 1955 Language in Rel ation to a Un i fi ed Th eo ry of the Structure of Hum an Beh avior, part II. Summer Institute of Linguistics, Gl endale, California. 1956 To wards a Th eo ry of the Struc ture of Human Be h avi 0 r. Sobret iro de Estudio s An tropologi cos, Mexico.

Pilling, Arnold R. n. d. Ch an ges in Ti wi Language. Wayne State University. MS.

P ittman , Ri chard ( Ed. ) and Kerr, Harl and 1964 Papers on the Languages of the Au stral ian Abori g ines. Occasional Papers in Aboriginal Studies No .3, Au stralian Institute of Aboriginal Studies, Canb erra.

Sommer, B. A. and E. G . n. d. Kunj en Phonology: Wo rd, Sy l labl e and Phoneme. Summer Insti tute of Lingui stics ( to be published) . 52

Strehlow, T . G . H . 1942 Aranda Grammar. Oceania, vol. XIII, No .1, Sep t ember 1942.

Ti wi (or Wo runuwi ) Gramm ar. MS. ( author unknown) .

B. A. an d E. G. SOMMER

O . Introduction. 1. Pronouns. 2. Social Organisation: Kinship. 3. Kinship Terms. 4. Correl at ion. 5. Re sidual Fo rms.

O. INTRODUCTION The fifty or sixty Kunjen (Oykangant) speakers who live at the Mitchell River Mission and on surrounding cattle stations are now largely detribalised, observing to a lesser and lesser degree the customs of their anc estors. 1 Thi s discussion of their pronominal and kinship terms brings to light a dist inct correlat ion between Kunj en linguistics and social behaviour.

1. PRONOUNS Kunj en pronoun s ar e free fo rms; the following matrix represents the Nominative fo rms:

Ch art Ia

Person Si ngular Du al Plural 1 incl. aliy amp ul excl. ay alinj anj tan NOMINA TI VE 2 inang upal urr 3 il ul etn

Kunj en wo rds fo llow a vowel-initial/consonant- final pat tern. 2 Comparison wi th the pronominal fo rms of geographic ally nearby languages indicates that conformity to this

Sommer, B.A. and Sommer, E.G. "Kunjen Pronouns and Kinship". In Glasgow, D., Glasgow, K., Kirton, J., Oates, W., Sommer, S.A. and Sommer, E.G. editors, Papers in Australian Linguistics No. 1. A-10:53-65. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-A10.53 ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. l

54 word-pattern has been achieved for Kunj en pronoun s largely by the loss of the initial consonant and/or final vowel. The Common Au stralian 1st pers. sing. nom. ngai suggested by Capell (1962) 3 , the Gugu-Yal anj i ngayu5 and the Wik Munkan ngay4 are thus reduced to the ay above. This system at ic change may be traced al so in the 3rd pers. dual nom. ul that has clear relat ionships with Gugu-Yal anj i pula, Koko Bera pulu and Wik-Munkan pul; the 2nd pers. plur. urr wh ich relates to Gugu-Yalanj i yurra, Koko Bera yurr; an d other forms. The persistence of pul in the dual forms of Kunj en is confirmatory evidence of its ant iquity and stability in the Cape York/Western Desert Languages, already remarked on at length by Capell (ibid. ). pul appears with a vowel change (Koko Bera yipel, 2nd pers. dual ), with loss of part of the form (ul above), minimally (Wik-Munkan pul, 3rd pers. dual ), or in inversion (Gugu-Yalanj i pula, Kunj en upal, 3rd per s. plur. ) . 6 The particle li noted by Capell app ears al so in the 1 st pers. dual fo rms; a feature shared by Gugu-Yalanj i and Koko Bera. These occurrences are significant in view of the very low level of coincidence of Kunj en wo rds with Capell 's "Common Austral ian" vocabulary (ibid. , 80ff. ). Only five Kunj en wo rds approach this CA list: el 'eye ' , al 'fire', uW'give', in :sitl , utj ir 'two'. Chart Ib indicates that the general marker fo r possessive case is -ang, and Chart Ic indicates -ngan as the general marker of the obj ective case.

Chart Ib

Person Si ngula r Du al Plural 1 incl. al ing amp ung excl. aten alinj ang anj tang POSSESSI VE 2 inin upang urrang 3 ingin ulang etnang

Chart Ie

Person Singul ar Dual Plural 1 incl. al ingan ampungan excl. atun al inj an anj tangan OBJECTI VE 2 inun upangan urrangan 3 ingun ulangan etnangan 55

The regul arity of fo rmation of the non- s ingular fo rms contrasts wi th the unpredictable singular fo rms.

2. SO CI AL ORGANISATION : KINSHIP There has been consi derable anthropological research into the social organisat ion of the Cape York tribes. R.M. and C. H. Berndt (1964, 80-81), commenting on this work, refer to Sharp (1939): 7 "The question of status is significant here, as in all Aboriginal societies to a lesser or greater degree, and thi sis expressed through sp eci fic status relationships. Sharp sp eaks of 'weak relat ionships' balanced by 'strong rel ationships', saying that this 'is necessary fo r a society in which every active individual relationship, at least between males, involves a definite and ac cepted superiori ty and inferiority. .. Even in distant relat ionships there is alway s present a recogn ized element of superordination and sub ordination ' (ibid. , 419). Th is inequal ity, in herent in the kinship terminology, is demonstrated in kinship behaviour, and is fur ther exemplified in asymmetrical or un ilateral preferential mar riage. " This concept of 'sup erordinat ion ' and 'subordination ', applied to kinship terminology in Kunj en, reveal s an inter esting structure. Since the marriages follow the general Murngin pattern, and the society is strongly patrilineal, males take priority over femal es where otherwi se any equal ity might be assumed. Membership in a senior gener ation confers superordinat ion. The numeral s in the following chart indicate a di fference in generat ion levels.

3. KINSHIP TERM S

Ch art II

Sup ero rdinate Subordinate apmingarr ff/mm arrngkapmalk ss/dd 2 atj ingarr fm/mf arrngkatjalk sd/ ds

(Continued on page 56) 56

Chart II - continued from page 55

Supero rdinate Su bo rdi na te ipangarr f/ fb amangarr m/ms arrngk child alangarr mb/ fsh uwangarr niece/ injangarr fs/mbw antangarr l nephew

olangarr ob akangarr yb epangarr os elangarr ys 0 orangarr h urrangarr w ulangarr cousin etnangarr cousin

Differentiat ion on the basis of generation and sex re mains valid up to the final " 0 " sector wh ere the upper series {orangarr, urrangarr} is superordinate to the lower { ulan�arr, etnangarr} , fo llowing the pattern of the preced ing "0 " sector ( brother/ sister) but the reason is less obvious. It is to be fo und in the status of the progenitors of ulangarr and etnangarr, certainly, but an established pattern has yet to develop.

4. CORRELATION A large proportion of the kinship terms set out above is derived from the possessive pronouns of Chart Ib, usually by the addition of -arr, the origin of wh ich is uncertain. urrang 'yours { plur. } ', urrangarr 'wife'; ulang 'theirs { dual } ', ulangarr 'cousin'; etnang 'theirs { plur. } ' , etnangarr 'cousin'; anjtang 'ours { excl. plur. } ', antangarr 'nephew' . Minimal change of a vowel in certain of the pronominal forms of Char t Ib has enabled the use of one basic form in expressing several relationships. ulang ' theirs { dual } '; ulangarr 'cousin'; thus modifies to olangarr 'older brother ', elangarr 'younger sister' and alangarr 'uncle '. The first per son pronouns { with the exception of anjtang 'ours { excl. plur. } '; antangarr 'nephew' } are not used to express kinship. 8 Th e reason fo r this is uncertain, but may perhaps involve the fact that kinship terms already express relat ionship s based on EGO . The failure to utilise the first person has made necessary the economy no ted above in minimal vowel changes. With the ex ception of some of these minimally derived forms, a further patterning may be evidenced by Char t II. 57

The singul ar and dual forms of Chart Ib are found in the 'sup erordinate' column, wh ile subordinat ion is expressed through plural fo rms.

5. RESIDUAL FORMS Wh ile injangarr 'aunty' is traceable to inin9 'yours (sing. )', terms such as amangarr 'mother ', akangarr 'younger brother ' cannot be traced directly to pronominal forms. It is suggested rather that the terms still unaccounted for may be derived from the wo rds fo r 'man' , 'child', and 'give'. pama '(native) man ' is used by the Kandju, Gugu- Yimidir, and Gugu-Yalanj i, and may be considered the original form of the Kunj en apm, wh ich has undergone vowel loss and inversion to conform to Kunj en phonology. It may be thi s fo rm, further modified by vowel change and/or consonant loss that is behind amangarr 'mother ' and ipangarr 'father '. This is supported by the occurrence of apm in the bloc indicating a generation difference of 2, wh ere it apparently denotes matrilineal or patrilineal continui ty of line. atj con trasts wi th apm in the same bloc; it occurs in isolation with the meaning of 'mud' . Berndts (1964) , in di scussing the responsibilities and privileges of the mother 's brother and father's sister, sug gest that these relat ionships are "p ivotal " and "crucial" in Aboriginal so ciety. These strong words express the deep interest anthropologists have shown in the mo ther 's brother/ father 's sister: nephew/niece relat ionship. The exceptional use of anj tang 'ours (excl. plur. )', antangarr 'nephew' has already been noted. uwangan, our closest dictionary entry to uwangarr 'niece ', has the meaning of 'might give ', wh ich, beside being a dep ar ture from the general pattern, carries interesting suggestions. 10 arrngk 'child' may be the result of the process of vowel loss and inversion app lied to akangarr 'younger brother '. Neither pul nor li can be clearly accounted fo r in any of the kinship terms. NOTES

1. Kunj en is a name popul arly app lied to the Oykangant, Olkol, Ayan and Winkan dialect speakers. 2. "Kunj en Phonology: Wo rd, Syllable and Phoneme" (forth coming) .

3. Capell, A. , "A New Approach to Au stralian Li"ngui stics", O Cleania Lin.guis tic Monograph No,, 1' ('1'9 6'2 )' Un iversity of Sydney. 4. Godfrey, Mar ie and Kerr, H. B. , "Personal Pronoun s in WiK-Munkan ", Papers on th e Languages of the Au stralian Abori gines (1964) A. I. A.S. Canb erra. 5. Oates, William and Lynette, "Gugu-Yalanj i Linguistic and Anthropological Data", Gu gu- Yal anj i an d Wi k-Munkan Lan gu age Studi es (1964) A. I. A.S. Canberra. (The bi-labial stop phoneme has been re-written as p for comp arative purposes. ) 6. The fo rm ngupal may al so be postulated as a proto- form for upal .

7. Berndt, R. M. & C. H. , Th e Wo rld of the Fi rst Au stralians, Ure Smith (1964) p. 509. Sharp, R. L. , "Tribes and Totemi sm in No rth-East Aus tralia", Oceania, Vol. IX, No s. 3 and 4, 8. Unless it is ampung 'our s (incl , plur. )' wh ich would re qu ire a complete change in the manner of arti culat ion of the mp clust er to w, and is less appealing than the suggestion above. 9. Schmidt's Group II Pronouns, wh ich Capell infers would apply to the We stern Desert/Cape York languages, includes njin, wh ich is strongly suggestive of a fo rm that would permi t this derivat ion of the kinship term. 10. Direct- address fo rms of the kinship terms ar e achieved in general by reduplicat ion (partial or entire ) of the first syllable of the usual reference term, and the loss of the final -arr. Thus ep angarr 'older sister' becomes

58 59

epepang; etnangarr 'cousin' become s netnang. The fo rms eparrntj (arr) ) etnarrntj (arr) ) etc. , wo uld be used to refer to 'my older sister', 'my cousin', etc. In discussions over a third person, the speaker honours the hearer by using the term denoting the hearer 's relationship to that third person. Anthropological research into the social organisa tion of the Kunj en people will no doubt afford more precise definitions of some relationship terms.

Acknowl edgmen ts

We are �ratefu l fo r the consultant help of members of the Summe r Instit ute of Lin�uistics in the wri tin� of this paper, and to Mrs El izabeth Hen ry B of the Mitchell River Mission wh o wi llin�ly helped with her lan�ua�e . Ortho�raP hy used is that establ ished by uOceaniau usa�e except that rr denotes a voiceless trill. Research into the Ku njen lan�ua�e by the authors has been supported by the Austral ian In stitute of Abori�inal Studies, Canberra .

* Sommer, B.A. and Sommer, E.G. "Kunjen Pronouns and Kinship". In Glasgow, D., Glasgow, K., Kirton, J., Oates, W., Sommer, S.A. and Sommer, E.G. editors, Papers in Australian Linguistics No. 1. A-10:53-65. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-A10.53 ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative.