This manuscript has been reproduced from the microfilm master. UM l films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type o f computer printer.
The quality of this reproduction is dependent upon the quality o f the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely afreet reproduction.
In the unlikely event that the author did not send UME a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, b^inning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back o f the book.
Photographs included in the original manuscript have been reproduced xerographically in this copy, ffigher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UM l directly to order.
UMl
A Bell & Howell Infoimation Company
300 North Zeeb Road, Ann Arbor M I 48106-1346 USA 313/761-4700 800/521-0600
Susan Ann Fitzgerald B.A.. U niversity of V ictoria. 1989 VI.A.. U niversity of Victoria. 1992
.A. D issertation S u bm itted in P artial Fulfillment of the R equirem ents for th e Degree of
D O C TO R OF PHILOSOPHY in the D epartm ent of Linguistics VVe accept this dissertation as conforming
to th e required standard
•. G.N. 0 Gradv. S u perviÊ i^(1
Dr. G.N. 0 'G rady. Supervifii^ (D epartm ent of Linguistics)
Dr. B .F. C < f ^ n . D epartm ental M em ber (D epartm ent of Linguistics)
Dr. T.VI. H e s ^ D ^ a rtm e n tc
Dr. T.VI. Hess ^ D ep art m ent al VIember (D epartm ent of Linguistics)
Dr. VV.H. A lkireyO utside VIember (D epartm ent of .Anthropology)
Dr. K.L. Hale, E xternal Exam iner (D epartm ent of Linguistics. Vlassachusetts In stitu te of Technology)
© Susan .Ann Fitzgerald, 1997 U niversity of Victoria
All rights reserved. This dissertation may not be reproduced in whole or in p a rt, by photocopy or other m eans, w ithout the perm ission of th e author.
Supervisor: Dr. Geoffrey N. 0 G rady
A B STRA C T
One of the most im portant questions in A ustralian com parative linguistics over the last 40 years is the validity of a Pam a-N yungan node in the A u stralian fam ily tree. Much of the com parative research done on A ustralian languages has su pported the notion of a Pam a-N yungan family, and its validity is now w ell-established. How ever. much work rem ains to be done, both in establishing the relationships am ong the Pam a-Nyungan languages and in reconstructing proto-Pam a-N yungan an d d e te r m ining the details of its developm ent in the various branches of th e fam ily tree. This dissertation is a contribution towards th e la tte r effort.
The prim ary purpose of the present study is to determ ine th e developm ent of the three initial velars. *k. *ng and *w, in 25 Pam a-N yungan languages through 1561 cognate sets. The cognate sets are also an im portant resource for th e stu d y of other aspects of phonological change in Pam a-N yungan languages. T h e d a ta provide evidence for the weakening of medial consonants, th e assim ilation of in itial velar glides and nasals to the following vowel, prenasalization of medial stops, th e developm ent of triconsonantal clusters, and th e presence of both a laminal lateral and a retroflex series of consonants in proto-Pam a-N yungan. In addition, statistical evidence is presented which supports the hypothesis th at assim ilation of the second to th e first vowel is an im portant process in the history of m an y Pam a-N yungan languages.
This dissertation also discusses im p o rtan t issues regarding the N eogram m arian hypothesis and the com parative m ethod. In particular, th e d a ta presented here sup port the idea th a t not all sound changes apply in a lexically ab ru p t, regular m anner. Many of the sound changes seen in th e d a ta appear to affect only a portion of th e eligi ble forms, and thus provide evidence for the theory of lexical diffusion. Furtherm ore, m ost of the changes are found not ju st in individual languages, b u t in a num ber of the languages under study. The d a ta therefore support the notion of pandem ic irregularity.
Examiners:
Dr. G.N. 0 'G rady. Sup^ndsor (D epartm ent of Linguistics)
---Dr. B .F. CarisQfl. D ep artm en tal M em ber (D epartm ent of Linguistics)
Dr. T.M . Hess. Dgpa&tmental M em ber (D epartm ent of Linguistics)
Dr. W ,H. Alkire.TC^utside M em ber (D epartm ent of A nthropology)
Dr. K.L. Hale. E xternal E xam iner (D epartm ent of Linguistics. M assachusetts In stitu te of Technology)
C o n te n ts
C on ten ts iv
L ist o f T ables v iii
L ist o f M aps ix
L ist o f A b b reviation s x
A ckn ow led gem ents xii
D ed ica tio n xiv
1 In tro d u ctio n 1 1.1 Previous work on P a m a -N y u n g a n ... 3 1.1.1 P h o n o lo g y ... 6 1.1.2 G r a m m a r ... 8 1.1.3 Lexicon ... 9 1.1.4 S e m a n t i c s ... 10 1.2 P ro to -P a m a-N y u n g a n ... 11
1.3 Languages in the S tu d y ... 14
1.3.1 Sociolinguistic In fo rm a tio n ... 16
1.3.2 D ata Sources and T r a n s c rip tio n ... 20
1.3.3 Phonem e In v e n to rie s ... 22
1.3.4 .\ssum ed S u b g r o u p in g ... 26
2 M eth o d o lo g y 29 2.1 Review of the Com parative M ethod ... 30
2.2 .Alternative Views on Sound C h a n g e ... 38
2.2.1 C a m p b e l l ... 39
2.2.2 Lexical D if f u s io n ... 41
2.2.3 L a b o v ... 43
2.2.4 Pandem ic Irregularity ... 46
2.2.5 Role of the .N'eogrammarian H y p o th e s is ... 48
2.3 Bottom -Cp and Top-Down C o m p a riso n ... 50
2.4 M ethod Used in this S tu d y ... 52
2.4.1 Overview ... 53
2.4.2 Details of the Method ... 54
3 *k -In itia l E tym a 63 3.1 "ka- ... 66
3.2 " k i-... 116
4 * n g -ln itia l E tym a 171
4.1 “n g a - ... 171
4.2 * n g i - ... 193
4.3 * n g u -... 19-5 5 * w -In itia l E tym a 206 •5.1 * \ v a - ... 206 5.2 *wi- ... 242 5.3 " w u - ... 254 6 A n a ly sis o f C ognate S e ts 262 6.1 Consonant C h a n g e s ... 262 6.1.1 Initial Velars ... 263 6.1.1.1 * k-> n g - ... 263 6.1.1.2 *k-> w - ... 265 6.1.1.3 >
0
... 267 6.1.1.4 *ng- > w - ... 268 6.1.1.5 *ng- > 0 ... 271 6.1.1.6 >0
... 272 6.1.1.7 S u m m a r y ... 273 6.1.2 Medial L e n itio n ... 275 6.1.2.1 Stops to N a s a l s ... 275 6.1.2.2 Stops to G l i d e s ... 276 6.1.2.3 Nasals to Glides ... 2766.1.3 .Assimilation of Initial Velars to Vi ... 277 6.1.4 P r e n a s a liz a tio n ... 278 6.1.5 Triconsonantal Sequences ... 280 6.1.6 Laminai L a te r a l ... 281 6.1.7 Retroflexes in proto-P a m a -N y u n g a n ... 283 6.2 .Assimilation of V? to V ; ... 285 6.2.0.1 Problems in R e c o n s tr u c tio n ... 286
6.2.0.2 Statistical Evidence for A s s im ila tio n ... 288
7 C onclusions 296
R eferen ces 300
L ist o f T ables
1.1 Dixon's proto-.Australian Consonant I n v e n to r y ... 12
1.2 O ’G rady's pPX Consonant I n v e n t o r y ... 13
1.3 List of S o u rces... 21
1.4 Phonem e Inventories (part 1)... ... 23
1.5 Phoneme Inventories (part 2)... 24
6.1 Observed Frequencies of and V'j ... 292
L ist o f M aps
L ist o f A b b r e v ia tio n s
BAA Bâgandji
BAY Bayungu
BGU Bidy ara- G ungabula BNJ Bundjalung DIY Diyari DYI Dyirbal GUM G um baynggir GUP G upapuyngu GYA Gugu-Yalanji KAU K aurna
KLY Kala Lagaw Ya XMA X garlum a XYA X yangum arta
XYA-S X yangum arta-Strelley dialect XYA-W X yangum arta-W allal dialect
XYU Xyungar
FIX Pintupi
PIT P itta P itta
PXK Pankarla
UMP Um pila
WEM Wembawemba
W JK W adjuk (X orthern Xyungar) W LP W arlpiri
VVMK Wik M ungkan WO I W oiwurrung
YAX Yanyuwa
YDX Yidiny
pCK proto-Central-K arnic pK A N p ro to K an y a ra
pK A R proto-K arnic pNG proto-Ngayarda pXY proto-Nyungic
pXY (D) proto-Nyungic (Desert languages) pNYY proto-Nyungo-YuuIngic
pP proto-Pam ic pPM proto-Pam a-M arie pPN proto- Pam a- Nyungan
pPN (E) proto-Pam a-N yungan (Eastern languages)
pPN - proto-Pam a-N yungan (no southeastern languages) pVVK proto-W estern-K arnic
A c k n o w le d g e m e n ts
I would like to th an k my supervisor. Dr. Geoffrey 0 Grady, for th e training he has given me in historical linguistics, for sharing his incredible knowledge of .Aus tralian languages, and for his unbounded enthusicism for my work. T hanks also to Dr. O"Grady for staying on as m y supervisor for many years after his retirem en t, and to Mrs. .Alix 0 ‘G rady for her support.
I would also like to thank th e o th er m em bers of my com m ittee. Dr. B arry Carlson. Dr. Thomas Hess and Dr. W illiam .Alkire. and my external exam iner. Dr. K enneth Hale, for th eir com m ents and suggestions. My thanks especially to Dr. Carlson for m any interesting courses and discussions, and for his confidence in me. Special thanks to Dr. Hale for going to the trouble and expense of attending my defense in person, and for his suggestions regarding future work.
Thanks to my fellow graduate students in the departm ent. My thanks especially to .Anne Gilbody and M urray Schellenberg for friendship, support and innum ber- able cups of tea. and to Violet Bianco, who has been both a kindred spirit and an inspiration to me.
I would like to acknowledge th e University of Victoria and the Social Sciences and Hum anities Research Council of C anada, for providing me with fellowships during th e course of my g raduate studies.
Finally. I would like to th an k my husband, K evin C attell. W ithout his help and com puter expertise, I would not have been able to u ndertake a study of this size. More im portantly, he has endured many ups and downs during our long years as students, and has given me his patience, wisdom, encouragem ent and love.
D e d ic a tio n
To my son. Jasper Cattell. and to my baby due in eight weeks'
time. You have taught me more about life than I could have
In tr o d u c tio n
O ne of the most im portant questions in A ustralian com parative linguistics over the last 40 years is th e validity of a Pam a-N yungan node in the A ustralian phylic tree. In 1962. Hale proposed that the Pam a-N yungan languages form a family, which is alm ost certainly a m em ber of the .A.ustralian phylum . A lthough m any researchers have agreed w ith this. Dixon (1980) felt th at the distinction between Pam a-Nyungan and non- P am a-N yungan languages is typological ra th er th an genetic. Thus, he believed th a t th ere is ju st one language family in .Australia, and th a t although the Pam a-N yungan languages are members, they do not share a period of common development sep arate from th e non-Pam a-Nyungan languages.
Much of th e com parative research done on these languages has supported the notion of a Pam a-Nyungan family, and its validity is now well-established. However, m uch work rem ains to be done, both in establishing th e relationships am ong the Pam a-N yungan languages and in reconstructing proto-Pam a-N yungan (pPN ) and determ ining th e details of its development in th e various brzinches of the family tree. T his dissertation is a contribution towards th e la tte r effort.
I also included roots beginning with w in eight languages which m ay weaken initial *ng to w. T he result of the study was 168 cognate sets which appeared to reflect n^-initial roots, either at the family level or at subgroup levels. These d a ta did not provide justification for establishing a regular conditioning environment for the weakening of
"^ng-.
The primary purpose of the present study is to determ ine th e development of the three initial velars. *k. *ng and *w. in 25 Pam a-N yungan languages through a large number of cognate sets. In addition, this study uses th e sets in exploring other issues in Pam a-Nyungan historical phonologv’. such as weakening of medial consonants, assimilation of initial velars to *i. prenasalization of m edial stops, the development of triconsonantal clusters, th e history of the lam inal lateral and of the retroflex series of consonants, and assim ilation of the second vowel to th e first. The sets are also useful as a source of corroboration for semantic links between putative cognates.
.Although the focus of this study is pP N initial velars, many of the reconstructions presented in Chapters 3-5 are for lower levels in the family tree. These data. then, may be useful for further work on subgroupings in the Pam a-N yungan family.
In the remainder of this chapter, 1 review past work on com parative Pama- Nyungan. provide a brief introduction to pPN and discuss th e languages included in this study. In C hapter 2 . 1 explore current issues regarding the com parative m ethod and the regularity hypothesis, an d discuss the applicability of the m ethod to Pam a-Nyungan languages. In addition, I detail the methodology used in this study. C hapters 3, 4 and 5 present th e cognate sets resulting from my study. T he develop m ent of initial velars, as well as other phonological questions, are discussed in C hapter 6. and my conclusions are given in C h ap ter 7.
1.1
P rev io u s w ork o n P a m a -N y u n g a n
In this section. I sum m arize th e work which has been u ndertaken to validate Pam a- Xyungan as a language family, to establish the subgroupings of the family, an d to reconstruct pPN . First, 1 discuss th e history of th e notion of Pam a-N yungan as a family. 1 th en list many of th e works which have appeared in the last 33-40 years.
Capell (1956:3) recognized the typological distinction betw een th e “prefixing" lan guages of m uch of the north and the “suffixing"’ languages of th e south and Q ueens land. In the former, prefixes are used to mark num ber and person on verbs. T h e la tte r lack prefixes entirely. Capell believed th at all of th e aboriginal languages of .\u stra lia are ultim ately related, but th a t too little evidence rem ains to reco n stru ct a proto-language on par with proto-lndo-European. Instead, he (1956:3) proposed th e notion of Com m on .A.ustralian. “a group of common elem ents, classifiable u n d er the headings of phonetics, morphology and vocabulary." C apell (1979) expands on th e history of th e .Australian languages.
A large-scale lexicostatistical stu d y of A ustralian languages undertaken in th e
1960s resulted in a prelim inary genetic classification which recognized 29 language families (O ’Grady. Voegelin and Voegelin (1966) an d O ’Grady. VVurm an d Hale (1966)). T he largest of these. Pam a-N yungan. covers approxim ately four-fifths of the continent and has approxim ately 170 daughter languages. T he northern b o u n d ary of this family roughly corresponds to the boundary between the prefixing and suffixing languages. The rem aining families are located in .Arnhem Land, th e K im berley region and part of th e G ulf country. This classification has been criticized by P la tt (1967) and by Dixon (1980), who both seem to m isunderstand its purpose. Dixon (1980:262-263) complains th a t th e comparison of th e vocabularies
language Y?’ was always provided w ith an answer ’yes' or 'n o .' . . . No criteria other th a n lexicostatistics were given for these [genetic] groups. . . . There was no a tte m p t to establish genetic relationship in the established sense . . . nor was there any adm ission th a t this was necessary.
Dixon (1980:262) also takes issue with th e fact th a t “in most cases only th e vocabu laries of adjacent languages were com pared.’’
O 'G rad y and Klokeid (1969:311) em phasize th a t “[tjhe purpose of th e creators of the classification was to apply a single technique - th e lexicostatistic one - to the task of arriving at a prelim inary genetic grouping of Australian languages.’’ They defend th e use of lexicostatistics, citing “th e lack of tim e to carry out a classifica tion based, for exam ple, on shared innovations in phonology" (O ’G rady an d Klokeid 1969:298). The classification was done under th e assum ption “th a t pairs of .Aus tralian languages which are geographically rem ote would turn out, in general, to be linguistically far apart also. . . . System atic spot checking in general su p p o rted the assum ption" (O ’G rady and Klokeid 1969:299). This allowed th e num ber of lexical comparisons to be greatly reduced, thereby m aking th e study possible.
D espite the criticism of P la tt and Dixon, m any A ustralianists see the cleissification as a useful starting point for further work. In fact, Blake and Dixon (1991:6) say th a t “it did provide a good overall picture of th e p a tte rn s of degrees of relatedness. ” Since its initial appearance, the classification has been altered to some extent; in most cases revisions are based on gram m atical and phonological evidence. U pdated versions of the classification ap p ear in W urm (1971 and 1972). .As well, som e studies of subgroups have indicated revisions. For exam ple, Yauyuwa was first classified as a non-Pam a-N yungan language, but is now considered to be Pam a-N yungan (Blake 1990); the Tangkan languages were classified as Pam a-N yungan, but have since been found to form a separate fam ily (Evans 1988:91); and the num ber of non-Pam
a-N'yungan families can be reduced. Such fine-tuning is only to be expected as more and more work on th e history of the A ustralian languages is undertaken.
Dixon (1980:226) also doubts w hether Pam a-N yungan is in fact a distinct node on the .Australian family tree. He says th a t “th e division betw een PN and nonPN languages is a typological (and areal) one. It should not be inferred that PN is in any sense a genetic unity - that there was a proto-PN . as an early descendant of pA [proto-.Australian].” He has since softened his position som ew hat, and in 1990 said th at Pam a-N yungan “m ight well be” a genetic entity, “but ad eq u ate justification has yet to be produced" (Dixon 1990b:400). By the next year. Blake and Dixon (1991:22-26) speak of Pama-Nyungan as a language family.^
In his critique of th e notion of Pam a-N yungan as a language family. Dixon (1980:221) states the steps necessary in establishing relationships am ong languages:
In order to prove th at a group of languages is genetically related , it is nec essary to undertake system atic com parison of their phonological, gram m atical and lexical systems: to put forward - on the basis of this com parison - a hypothesis about some of th e forms and stru ctu res that can be assigned to their putative com m on ancestor; and th en to detail the regular changes by which each m odern language has developed from the proto-system .
In the rem ainder of this section, 1 discuss m any of the studies which have been published on Pam a-N yungan in the last 35 years.^ The list of such works shows
^Note that Blake has challenged Dixon’s views on Pam a-Nyungan, and his work on pronouns (1988. 1990) has been instrum ental in establishing the genetic unity of Pama-Nyungan and its relative genetic distance from the non-Pama-Nyungan languages.
'I have attem pted to include as many works in this discussion as possible. Unfortunately, due to tim e and resource lim itations, I may have om itted some applicable studies. No offense is intended by such omissions, and I would appreciate these being brought to my attention.
th at the steps outlined by Dixon are all in progress. T h e com plete establishm ent of a language fam ily entails a huge am ount of work, m ore th a n can be done by any one person or sm all group of researchers, and m ore th a n can be done in a short am ount of tim e. As can be seen below, “system atic com parison of [the] phonological, gram m atical and lexical system s" of Pam a-N yungan languages is being undertaken, and “hypothes[e]s about some of the forms and stru ctu res th a t can be assigned to their putative com m on ancestor" are being put forward.
1.1.1
P h o n o lo g y
Since Hale’s stu d y of the .Grandie languages in 1962, m any papers on the historical phonology of the Pam a-N yungan languages have appeared. These include studies of th e development of dialects of a single language, of groups and subgroups of the fam ily and of the family as a whole, as well as proposals regarding th e phoneme inventory and phonotactics of pPN.
Hale (1964) gives a phonem e inventory for proto-P am ic. and dem onstrates th a t 13 dialects and languages of th e N orthern Pamic subgroup, which were at one tim e thought to be “u n -.Australian," are indeed mem bers of th e Pam a-N yungan family. O 'G rady (1966) presents a detailed study of proto-N gayarda phonology and of th e development of th e Ngayarda languages.
Dixon (1970) discusses th e developm ent of lam inal stops and nasals in A ustralian (mainly Pam a-N yungan) languages. Alpher (1972) looks at th e languages of so u th western Cape York Peninsula. In 1976, a large num ber of papers were published on the historical phonology of Pam a-N yungan languages. T hese include Alpher (1976) on innovations found in Cape York Peninsula languages; Crowley (1976) on the relation ships of seven languages of New England, paying p a rtic u la r atten tio n to th e phono logical changes found in Nganjaywana; Hale (1976b, 1976c, 1976d) on the changes in
several N orthern Pam ic languages, in Uradhi. an d in th e W ik languages, respectively, from proto-Pam ic; O G rady (1976) on the developm ent of th e U m pila phonem e inven tory from th at o f proto-Pam ic: Rigsby (1976) on K uku-T haypan. a Pam ic language: Som m er (1976b) on m etathesis and the loss of in itial consonants in three languages of Cape York P eninsula; and Sutton (1976a) on several initial-dropping languages of southern C ap e York Peninsula. Blake (1979b) discusses th e phonological history of Kalkatungu, while Hercus (1979) exam ines th e dropping of in itial consonants in .\rabana-W angkangurru. O 'G rady (1979) proposes a phonem e inventory for proto- N uclear-Pam a-N yungan.
Black (1980) studies th e Norman Pam ic subgroup. Dixon (1980) includes sections on proto-A ustralian phonology and phonological change: given th a t his work involves m ainly Pam a-N yungan languages, his "proto-A ustralian" can be tak en to be pPN . .Austin (1981b. 1988) details the developm ent of th e K anyara an d M an th arta lan guages of W estern .Australia. O 'G rady (1987) discusses phonological changes result ing in Eastern Pam a-N yungan monosyllabic roots, and Evans (1988) is an im portant work on the th e m erger of proto-.\u stralian apicals w ith lam inais in pPN .
Interest in th e com parative phonology of Pam a-N yungan has not waned in the present decade. A ustin (1990) discusses the developm ent of th e K arnic group. Dench (1990) is an au tosegm ental analysis of ap parent diachronic m etath esis in certain di alects of Nyungar. Dixon (1990a) details phonological change in dialects of Dyirbal. and Hendrie (1990) studies initial apicals in nine P am a-N yungan languages. O ’G rady (1990e) looks a t th e phenomenon of prenasalization in Pam a-N yungan languages. Fitzgerald (1991) studies th e reflexes of pP.N *ng- in 20 languages, while Chen (1992) exam ines the developm ent of pPN *w- in the sam e languages. S h arp e ( 1993) discusses th e loss of final r in Bundjalang. Blake and R eid (1994) look at sound change in the Kulin languages of central and western V ictoria, as do Blevins an d M arm ion (1994) in N hanta. O ’G rad y an d Fitzgerald (1995) discuss triconsonantal sequences in a num ber
of Pam a-N yungan languages, and A ustin (1997) outlines th e phonological develop ment of the Icmguages of central New South Wales. Breen (1997) discusses rhotics and apical stops, while Crowley (1997) looks at th e use of th e com parative m ethod in .Australia, focussing on the languages of northern New South Wales. Fitzgerald ( 1997) presents a prelim inary analysis of the history of th e lam inal lateral. Koch ( 1997) looks at the phonological developm ent of th e Arandic languages, spoken in the southeast of the N orthern Territory. S harpe (1997) exam ines th e dialects of Bundjalung.
1.1.2
G ram m ar
Many of the above-m entioned works also include in th eir scope the diachronic study of morphological and syntactic system s. These studies include O 'G rady (1966), .Alpher (1972). Hale (1976b. 1976c), Blake (1979b), Black (1980) and A ustin (1988). In addition. Hale (1970) looks at passive and ergative constructions. Case m arking is a well-covered topic: studies include Hale (1976a) on ergative and locative suffixes: Sommer (1976a) on ergative and instrum ental case m arking in central Cape York Peninsula: Blalce (1977. 1979a): and Klokeid (1978) on th e developm ent of accusative and ergative system s in Pam a-N yungan languages. .As well. Dixon (1977) discusses syntactic change in .Australian languages.
Dixon (1980) briefly discusses th e antipassive construction and the development of accusative case m arking system s. O 'G rady (1981a) provides a reconstruction of the first person dual inclusive pronoun in pPN , as well as a suggestion regarding its origin. Dench (1982) details th e developm ent of accusative-type case m arking in the Ngayarda languages. Nash (1982) looks at W arlpiri verbs and preverbs. Blake (1987) discusses th e developm ent of system s of m arking core functions, while Bavin and Shopen (1987) exam ine changes occurring in th e W arlpiri pronom inal system and Blake (1988) reconstructs pP N pronouns.
Alpher (1990) studies pPN inflectional verb morphology, an d Blake (1990) pro vides a further look at the use of pronouns in the genetic classification of A ustralian languages. Heath (1990) com pares verbal inflections in Pam a-N yungan and non- Pam a-Xyungan languages. Nash (1992) examines the kinship affix *-rti: Dench ( 1994a) details changes affecting pronoun paradigm s of 13 P ilb a ra region languages: and Koch ( 1996) presents several case studies of morphological change in Pama- Xyungan languages. Alpher (1997) conducts a prelim inary stu d y of possible morpho logical sources of noun-final n in Cape York languages. Dench (1997) explores the morphological development of complex kinship term s, and Hercus (1997) surveys the use of the verbalizer -ma- in th e form ation of adverbs.
1 .1 .3
L exicon
Capell began the work on lexical reconstruction in 1956 with his Com m on .Australian word list. Many of the com parative Pam a-N yungan studies which have been un dertaken since then have either focussed on or included lexical reconstructions. For exam ple. Hale (1964. 1976b. 1976c. 1976d) reconstructs proto-P am ic roots. O"Grady (1966) presents 451 reconstructed stem s for proto-Ngayarda, w ith some discussion on further reconstructability to various genetic levels. Merlan (1979) focuses on verbs, and O ’Grady (1979) gives reconstructions at the proto-N uclear-Pam a-N y ungan level and at various sublevels.
Black (1980) presents reconstructions for proto-Norm an Pam ic. Dixon (1980) includes a num ber of lexical reconstructions which, although labelled as "proto- Australian,"' are more properly considered pPN . Austin (1981b) gives 475 reconstruc tions for proto-K anyara and proto-M ant hart a stem s. O ’G rady (1981a, 1981b. 1987) reconstructs assorted roots, including proto-N uclear-Pam a-N yungan *yamu. .\lpher and Nash (1982) study the ra te of lexical replacem ent in A ustralian languages. Néish
(1982) focuses on the etym ology of W arlpiri kurdungurlu. Austin (1988) provides further reconstructions for proto-K anyara and p ro to -M an th arta.
A ustin (1990) contains lexical reconstructions for proto-K arnic and its subgroups. Blake and Reid (n.d.) present a large num ber of cognate sets for the languages of V ictoria. In th e 1990s. O 'G ra d y and his students have undertaken several stu d ies which include a large num ber of lexical reconstructions at various levels. Some of these, such as Hendrie (1990), O ’G rady (1990e), Fitzgerald (1991, 1997), Chen (1992) and O ’G rady and Fitzgerald (1995) were m entioned in Section 1.1.1. Fur ther such works include O ’G rady (1990b) on the reflexes of the universal quantifier
*pamtung in Pam a-X yungan languages, as well as on num erous other roots: O ’G rady
(1990c) on roots with *m-. *j- and *k-: O ’G rady (1990d) on roots beginning with
*pa-: and O ’G rady (1990f) on cognates shared by VVadjuk and Umpila; In addi
tion. A ustin (1997) provides 349 reconstructions for proto-C entral New South Wales. Evans and Jones (1997) present a num ber of cognate sets involving term s for wooden and stone im plem ents, while Koch (1997) extends m any published cognate sets to include .Arandic languages. McConvell (1997a) exam ines th e development of kinship term s in Ngumpin-Yapa. an d Nash (1997) studies flora terminology in a num ber of languages. O ’G rady and F itzgerald (1997) discuss searching for cognates in Pam a- Nyungan languages.
1 .1 .4
S em a n tics
Several studies have focused on or included discussion of sem antic change, both in Pam a-N yungan and in A ustralian languages in general. Schebeck (1978) sum m arizes associations between body-part term s and other concepts in the Yuulngu languages of north-east A rnhem Land, and Som m er (1978) discusses the sem antic associations of eye’ and 'no good.’ O ’G rady (1979) provides illu stratio n s of com m on sem antic
relationships in A ustralia, as well as discussion of unclassifiable relationships. Dixon (1980) exam ines types of semantic shift, while Evans (1990) studies th ree “sem ajitic subnetw orks” in A ustralia. Evans (1992) discusses the use of synchronic polysem y in studying sem antic change in A ustralia. W ilkins (1996) presents exam ples of n a tu ra l sem antic changes taken from .Australian languages, and Evans (1997) examines links between flora and fauna terminology. McConvell (1997b) explores th e association between Ash' zmd m e a t'. Sharpe (1997) looks at semantic shifts exemplified by th e Bundjalung dialects, while Wilkins (1997) studies cultural sources of evidence for sem antic associations.
Given th e num ber and variety of works dealing with Pam a-X yungan. it is clear th a t we are well on th e way to fulfilling the conditions for d em o n stratin g genetic re la t edness. Especially im portant are works such as Blake (1988). Evans (1988) and H eath
(1990) which com pare Pam a-X yungan and non-Pam a-X yungan languages, and pro vide evidence th a t Pam a-X yungan is indeed a distinct family (although probably u lti m ately related to th e non- Pama- X y ungan fam ilies). Unlike Dixon, m any researchers believe Pam a-X yungan to be "well-established by the com parative m ethod" (Ross and Durie 1996:33).
1.2
P ro to -P cu n a -N y u n g a n
McConvell ( 1996a: 125: 1996b:3)) proposes th a t pPN may have broken up around 6000 B.P. He arrives at this figure by "the m ethod of 'backtracking' [in which] known stages of expansion and cu ltural diffusion are sequenced making reasonable inferences a b o u t the tim e needed between each stage and ensuring th at sequences in different regions which link together also connect chronologically” (McConvell 1996a:125). Xichols (1992:211) states th a t "the Pam a-X yungan fam ily . . . is of com parable age” to Indo- European, which is about 6000 years old. O ’G rady (1979:111. 1990a:xii), however.
feels th a t the tim e depth for proto-Pam a-N yungan is som ew hat shallower, at approx im ately 4000 years. He bases his estim ate on th e am ount of change th at has occurred, which he believes is on par with th a t in the Finno-U gric fam ily ( 0 'Grady, p.c.). Blake and Dixon (1991:25) believe Pam a-N yungan to be “relatively young," although they do not posit a specific age.
Questions regarding th e origined hom eland of the Pam a-N yungan people, and their spread through the modern Pam a-N yungan axea have enjoyed renewed interest of late. The reader is referred to McConvell (1996a. 1996b) and Evans and Jones (1997). who present a variety of evidence relatin g to th e expansion of the Pam a- Nyungan family. Dixon (1980:158. 1990b:393) posits th e consonant inventory seen in Table 1.1 for proto-.Australian. Since Dixon’s work on com parative phonology has
Table 1.1: Dixon’s proto-.Australian Consonant Inventory labial apical lam inal velar
stops p t j k
nasals m n ny ng
laterals I (ly)
rhotics rr. r
glides (w)^ y w
focused almost exclusively on Pam a-N yungan languages, we can compare this table to O ’G rad y ’s (1979:131) proto-Nuclear Pam a-N yungan inventory. The only differences between the two are that O Grady tre ated r as a glide, and did not doubt the presence of a lam inal lateral.
In later work, O ’Grady (1990a:xxi) modified his pP N consonant inventory, given in Table 1.2. Although many Pam a-N yungan languages have b o th a palatal and a dental lam inal series, both Dixon and O ’G rady include ju st a single laminal series for th e proto-language, as posited by Dixon (1970) (b u t see Koch (1997)). P alatal symbols
Table 1.2: O ’G rady’s pPN Consonant Inventory labial alveolar retroflex lam inal velar
stops P t rt j k
nasals m n rn ny ng
laterals 1 rl ly
rhotics rr
glides (w) r y w
are used to represent this series. Both system s also contain a labial and velar series. In addition, Dixon and O 'G rady agree th at th e proto-language had three vowels ( i.
a and u). with distinctive length in the initial syllable.
The above inventories differ in three aspects. First, Dixon treats r as a rhotic. while in O 'G rad y ’s system it is a glide. Second. Dixon proposes a single apical series, which 0 Grady divides into an alveolar and retroflex series. T hird. O G rady posits a lam inal lateral ly. while Dixon is unsure of its presence. In C h ap ter 6. the latter two issues are exam ined in light of the present study.
Much about th e phonotactics of pPN has been determ ined ( 0 G rady I990a:xxii). Most roots were disyllabic, and nasals, laterals, rhotics and glides could occur word- finally. Stops were found in this position only in onom atopoetic interjections. .A.s is the case in most .Australian languages. pPN words were all consonant-initial. The distinction between th e alveolar and retroflex apical series was neutralized in initial position, and *rr was not found in this position. .Although O G rady (1990a:xxii) has stated that *ly was also disallowed word-initially, he has since reversed his position because of the recently noted presence of Ih in initial position in Yanyuwa (O ’Grady. p.c.). O ’Grady and Fitzgerald (1995) present evidence th a t, in addition to biconso- nantal medial sequences, pPN allowed triconsonantal m edial sequences m ade up of a non-naaal resonant followed by a homorganic nasaJ-stop cluster.
1.3
L anguages in t h e S tu d y
The 25 languages used in this stu d y are Bâgandji, Bayungu. B idyara-G ungabula. Bundjalung, Diyari. Dyirbal. G ugu-Y alanji. Gum baynggir, G upapuyngu. Guugu Y im idhirr. Kala Lagaw Ya. K aurna. Ngarluma. N yangum arta. N yungar (of which VVadjuk is a dialect). Pankarla. P in tu p i. P itta P itta , Um pila. W arlpiri. VVembawemba. VV’ik M ungkan. Woiwurrung, Yanyuwa and Yidiny. They were chosen for th eir rela tively conservative phonological system s and for th eir geographical spread across the Pam a-N yungan area; see Map 1 for th eir approxim ate locations.
SN J •
I I
Pama-Nyungan ^ 0 Non-Pama-NyunganI I
Papuan and AustronesianAffiliation Unknown
Map 1: Approximate Locations of Languages
BAA Baagandji NYU Nyungar
BAY Bayungu PIN Pintupi
BGU Bidyara-Gungabula PIT P itta-P itta
BNJ Bundjalung PNK Pankarla
DIY Diyari L-MP Umpila
DYI Dyirbal WEM Wembawemba
GUM Gumbaynggir W JK Wadjuk
GUP Gupapuyngu W LP Warlpiri
GYA Gugu-Yalanji W MK Wik-Mungkan
KAU ECauma WOI Woiwurrung
KLY Kala Lagaw Ya YAN Yanyuwa
NMA N'garluma YDN Yidiny
N ineteen of these languages (all but Gumbaynggir, Kala Lagaw Ya, N garlum a. Pankarla, W oiwurrung and Yanyuwa) were also used in Fitzgerald (1991) and Chen (1992): in this study. Bundjalung is used in place of G idhabal. which is one of its dialects.
In this section. 1 give brief sociolinguistic inform ation, sources of d a ta and the phonem e inventory for each language. I also discuss th e subgroupings of th e Pam a- Nyungan fam ily assumed for these languages.
1 .3 .1
S o cio lin g u istic In fo rm a tio n
The languages in this stu d y range from extinct, such as W oiwurrung, to relatively strong and viable, such as W arlpiri. M any of the languages have a sm all num ber of speakers left and are no longer being learned by children. O thers are the first languages of children in th eir com m unities a n d /o r are being used in school program s. In the following discussion, languages are listed in the subgroup order given in Section 1.3.4. 1 give th e location of each language, and I try to give up-to-date inform ation on th e num ber of speakers and of the statu s of the language in the com m unity: this has not been possible in all cases.
T here are approxim ately 800 speakers of N yangum arta (NYA) (Sharp 1997:329). which is spoken at Strelley (-S). LaGrange and Port Hedland in W estern .\u stra lia . It was spoken at Wallal (-W ) until the 1970s.
W arlpiri (W LP), with approxim ately 3000 speakers, is "one of th e strongest sur viving Aboriginal languages” (Bavin and Shopen 1991:104). It is th e com m unity language at Yuendumu, Willowa and Lajam anu in central A ustralia (Bavin 1993:85). and heis been used in bilingual program s (Yallop 1982:164).
P intupi (PIN) is a dialect of the W estern Desert language, which has over 3000 speakers. PIN is spoken at Balgo Hills an d Billiluna in W estern A ustralia, an d at Papunya and Yuendumu in th e N orthern T erritory, and has been used in bilingual program s (Yallop 1982:47-48. 164).
K aurna (KAU) was the language of th e .Adelaide Plains. It was recorded by Teichelmann and Schurm ann in 1840. and th e last fluent speaker died in 1929. KAU language programs at both th e elem entary and secondary levels have recently been undertaken ( .Amery and Simpson 1994:144-148).
Pankarla (PNK) was recorded by Schurm ann in 1844. It was noted by Hale an d O G rady in 1960 as having partial speakers (O ’Grady, p.c.)
Nyungar (NYU) is a collection of dialects which Wcis originally spoken in so u th western Western .Australia. Few speakers rem ain today; most of these speak th e eastern dialect (Dench 1994b:173-174). VVadjuk (W JK ) was th e northern dialect of Nyungar. and was spoken in th e area of P erth (Dench 1994b: 175). It was recorded in the nineteenth century by Moore.
Ngarlum a (NM.A). studied by Hale in 1960. probably has no fluent speakers re m aining today.
Bayungu (B.AY) was traditionally spoken in northw estern W estern .Australia, “along the coéist and inland along the M inilya and Lyndon Rivers north-east of th e present-day town of C arnarvon.” The few rem aining speakers of BAY live in C arn ar von (.Austin 1992:v).
Gupapuyngu (G U P) is a m em ber of th e Yuulngu subgroup of th e P am a-N yungan enclave in northeastern A rnhem Land. T h e Yuulngu languages have m ore than 3000 speakers. GUP is a m ajor dialect spoken a t M ilingimbi and has been used in bilingual
programs (Yallop 1982:54. 164).
Yanyuwa (Y.A.N) has approxim ately 100-200 speakers in th e area of Borroloola. Some members of th e younger generation have a passive knowledge of the language (M egan Morals, p.c.).
Kala Lagaw Ya (KLY) is spoken on th e W estern Torres S trait Islands. .\s of 1991. th ere were 3500-4000 speakers of K ala Lagaw languages on th e western islands (Ford and Ober 1991:118).
L’mpila (U M P) is spoken in northern C ape York P eninsula (Harris and O 'G rad y 1976:165).
Wik-Mungkan (W M K ) is spoken at A urukun on th e west coast of Cape York Peninsula, where speakers of m any W ik-dialects. as well as of o th er languages, moved from the 1920s to the 1960s. W M K has become th e m ain language spoken in th e community. It was also used in a bilingual school program from 1973 to 1988. and has recently been re- intro d u ced into th e school system (Sayers 1994:352-353). In
1982, WMK was estim ated to have 800-1000 speakers (Yallop 1982:44).
Guugu Y im idhirr (Y IM ) is a language of northern Q ueensland, and most of its present-day speakers live a t Hopevale Mission, north of Cooktow n. .As of 1978. Guugu Y im idhirr was still the first language of children in th e com m unity, although th ere were no bilingual school program s (H aviland 1979:27. 35).
Gugu-Yalanji (GYA) is spoken in southeastern C ape York Peninsula. In 1982. GYA had 500-600 speakers (H ershberger and H ershberger 1982:v). and had been used in bilingual program s (Yallop 1982:164).
Yidiny (YDN) was th e language of th e C aim s-Y arrabah area of Queensland. .As of 1991, only a few speakers were still alive (Dixon 1991:25).
D yirbal (DYI) was also spoken in the Cairns rain forest region of Queensland. In 1972. there were approxim ately 30 speaJcers, living mostly in the M urray Upper region (Dixon 1972:25. 37).
Bidyara-G ungabula (BG U ) was spoken at .\u g ath ella. Charleville and Clermont in southern Queensland. In 1972. a num ber of elderly speakers rem ained (Breen
1973:3-5).
B undjalung (BNJ) was spoken in northeastern New South Wales. Only a few. elderly native speakers are left (Sharpe 1994:1).
Gum baynggir (GUM) was also spoken in north ern New South Wales, and was a neighbour of Bundjalung. As of 1979. a single speaker remained (Eades 1979:244).
W embawemba (W EM ) was a Kulin language spoken in western V ictoria (Hercus 1994:100). Only elderly speakers were still alive in 1969 (Hercus 1969:9).
W oiwurrung (WOI) was th e traditional language of central V ictoria, including the M elbourne area. The language could not survive European settlem ent of the area, and by th e 1960s. only a salvage study could be undertaken. Fortunately, several word lists were recorded in th e late nineteenth century (Blake 1991:31-58).
Bâgandji (B A .\) was once spoken in western New South Wales, along the Darling River (Hercus 1976:229).
Diyari (DIY) was spoken in northern South A ustralia. .\t the tim e of first Eu ropean contact, there were probably over 1000 Diyari speakers; as of 1994. only two fluent speakers rem ained (A u stin 1994:125-126).
P itta -P itta (PIT ) was a dialect of an unnam ed southwestern Queensland language. In 1979, there were ju st two speakers left (Blake 1979c:183-l85).
1 .3 .2
D a ta S ou rces an d T ranscription
The word lists for the languages in this study are of widely varying length. The YIM vocabulary, for exam ple, is th ree and one-half pages long, while th e W LP dictionary, which includes a vast num ber of entries, derived forms and exam ple sentences, takes up over 6.5 megabytes of disk space. For WLP. NYU, BAY an d YDN. which were also used in Fitzgerald (1991), new sources of data have recently appeared. In addi tion. a dictionary for all th e Bundjalung dialects has also been recently published; it subsumes th e Gidhabal dictionary used in Fitzgerald (1991). T h e database for the present stu d y includes all velar- and vowel-initial forms. Table 1.3 lists the source(s) of d a ta and the num ber of item s in the database for each language.
For most of the languages, field work has been undertaken by m odern linguists: exceptions to this are K.A.U and PNK which were both recorded in the nineteenth century. Although these sources are not w ritten in a m odern transcription. Capell (1971:668) says th a t ”[t]he G erm an writers used a phonemic scrip t which is still quite a passable guide to the sounds intended." Much of the d a ta presented here for W.JK. a dialect of NYU. is also from the nineteenth century; however, m ore recently recorded d ata are also used. .Approximately 100 words were recorded for W OI in the 1960s by Luise Hercus (Blake 1991:58). Otherwise, d ata for this language are also from the nineteenth century. Blake (1991) heis standardized the tran scrip tio n s from th e old sources in light of Hercus' work and of “the fact th at A u stralian languages te n d to have very sim ilar sets of ... phonem es’’ (Blake 1991:58).
There is considerable variation in the transcription system s used by lexicographers of A ustralian languages. For exam ple, since relatively few languages have a voicing or fortis/lenis distinction, some sources use b, d and g, while o th ers use p, t and k. M any employ a practical orthography, which contains only ty p ew riter-ty p e symbols: thus, retroflexes are w ritten as rt. m and rf, g as ng, ? as and so on. O thers use symbols
Table L3: List of Sources
N yangum art a- S t relley Hale et al. (1980) 641 N yangum arta-W allal O 'G rady and F itzgerald (in prep.) 836
W arlpiri Laughren et al. (1995) 2273
Pintupi Hansen and Hansen (1974) 1015
K aurna Teichelmann and Schurm ann (1840) 544
Pankarla Schurmann (1844) 852
N yungar Douglas (1976) and Dench (1994) 591
VVadjuk Moore (1884 ) 681
N garlum a Hale (1982) 671
Bayungu O 'G rady s field notes 474
G upapuyngu Lawton and Lowe (n.d.) 1022
Yanyuwa Bradley et al. (1992) 1842
Kala Lagaw Ya Bani and Bani (1975) 870
Umpila O 'G rady (n.d.) 468
Wik M ungkan K ilham et al. (1986) 917
G uugu Y im idhirr Haviland (1979) 135
Gugu-Yalanji Hershberger and H ershberger (1982) 643
Yidiny Dixon (1991) 822
Dyirbal Dixon (1972) 224
B idyara-G ungabula Breen (1973) 316
B undjalang Sharpe (1992) 713
G um baynggir Eades (1979) 247
W embawemba Hercus (1969) 261
W oiwurrung Blake (1991) 438
Bâgandji Hercus (1982) 557
Diyari .\u s tin (1981a) 237
P itta P itta Blake (1979c) 311
from the phonetic alphabet and diacritics. In addition, som e of th e languages in this study, such as G U P. have well-established orthographies used by th e speakers.
In order to facilitate com parisons among the languages for th e purposes of this dissertation, m ost of th e d a ta are presented in th e "ptk" orthography. For stops, fricatives, nasals and laterals, th e following symbols are used: labials p and m: alveo- lars t. s, n and /: retroflexes rt, m and rh dentals th, nh and Ih: palatals j, ny and ly, velars k and ng; an d glottal stop ’. For those languages having a voicing or fortis/lenis
distinction, both voiceless ajid voiced sym bols are used: the cluster consisting of an alveolar nasal and a voiced velar stop is w ritten n=g to distinguish it from th e velar nasal. For the rhotics. r r is an alveolar fla p /trill, r is a retroflex glide and rrh is an alveolar trill in languages having b o th a flap and a trill. T here are two additional glides, y and w. Most of the .\u stra lia n languages have at most five phonem ic vowel qualities, i. e. a. o and u: length is in d icated by two vowels, ii. ee. aa. oo and uu. D ata from nineteenth-century sources are left unchanged because in m any cases we cannot be certain of the actual pronunciation. T hese d a ta are given in sm all u p p er case symbols to remind the reader that th ey are not w ritten in a stan d a rd tran scrip tio n system . Forms from W OI are given as in Blake (1991).
1.3.3
P h o n em e In v en to ries
The phoneme inventories for the languages in this study are given in Tables 1.4 and 1.5.-*
"*The caption “F” refers to fricatives. Some dialects of NYU have developed phonemic vowel length; see Dench (1994b:9-10). Breen (1997:75) provides possible alternative phonem e tables for BGU. The inventory seen here is the one given in the BGU d ata source (Breen 1973). Hercus (1994) does not use a or o for WEM. However, the d a ta in th is study are taken from Hercus (1969), which does use them.
= = = = - = ss z X 1 4 : 4 : 4 4 < < 4 < 4 4 i 4 % 4 5 :> X « - - — -J ÿ & 5 i 5 >v >% >4 > - > - >* > >» >k >v >» X X. X X z - - - -* 3 : z X XX X XX t c >» >k >• J£ .X X s — % — X X — “Z - - - z _2 — — — — — afl bC so ~ c qO Qfi Qfl - “ - z Z >k % > > z z z >• j -"3 z z z z - - “ -3 z E zX z E E E = = = z zz z z zz ? = z zz = = = = z z z _ ~ Z Z z " “ " X XX c j g X 3 c _4 9fl > - , > •-» - - *-5 J S .X X X C-% -- - - r- z t z - - g - - - - r -. X £ _ - a - — X 30 < — z — z < > • z > z < Z > s < < z 5 Z Z 5 z X >
>i se > v >l > i » > k % $ >% » > v & > v se % C > op = go X gp X iC z c: X zX S < < X > ?
G U P is th e only language in this stu d y w ith th e full set of seven place distinctions found in th e Pam a-N yungan family. Only G U P, W M K an d UM P have a g lo tta l stop: NYA. W LP and PIN have only one lam inal series: an d m ost eéistern languages lack a retroflex series. GYA. YDN. DYI. BN J and GUM show only four places of articu latio n for stops and nasals. Regarding YIM. H aviland (1979:37) states th a t "[t]he s ta tu s of the retroflex stop and nasal as distinct phonem es is som ew hat problem atic." KLY is one of the few .Australian languages w ith fricatives. A fortis/lenis distin ctio n has developed in GU P and KLY and in the apical series in DIY.
It is usually th e case th a t A ustralian languages have a nasal corresponding to each stop. Exceptions among th e languages in th is stu d y are KLY. BGU and W E M . each of which is “missing" at least one néisaJ.
Most of th e languages have either a lateral for every non-peripheral place of a r ticulation. or ju st a single lateral. Exceptions to these generalizations are G U P. WOI and possibly W JK (although the NYU d a ta indicate th a t W JK did have a retroflex lateral). Most of the languages also continue th e two pPN rhotics. although these have merged in KLY and BNJ. and pPN *r becam e a g lo ttal stop in UMP and W M K. T hree languages. WLP. P IT and DIY. have developed a th ird rhotic. W LP rd is an apico-postalveolar flap and PIT and DIY rrh is an apico-alveolar trill.
Most of th e languages in this stu d y have a three-vowel system : additional vowels have developed in NYU. W JK (based on NYL d a ta ), KLY, WMK. BN J, W E M . and possibly in K.AL\ PNK and WOI. The phonological inventories of W JK . K.ALT PNK and W OI cannot be stated with certainty, because all or m ost of th eir d a ta a re from nineteenth-century sources. For exam ple, it is not clear w hether W OI had one o r two lam inal series, a retroflex lateral, one or two rhotics, and th re e or m ore vowels (Blake
1.3.4
A s su m e d S u b grou p iu g
The following subgroupings are assumed in this dissertation:"
Xyungo-Yuulngic G roup Xyungic subgroup M arrngu sub-subgroup Yapa sub-subgroup W ati sub-subgroup Yura sub-subgroup X yungar sub-subgroup X gayarda sub-subgroup
K anyara-M antharta-Y uulngu subgroup K anyara sub-subgroup
Yuulngu sub-subgroup VV'arluvvaric G roup
Mabaygic G roup Pama-M aric G roup
M iddle-Pam a subgroup E astern-P am a subgroup A therton-P am a subgroup Yara subgroup Mari subgroup Bundjalungic G roup Gumbaynggiric G roup XYA W LP PIN GAW. PN K NYU. W JK XMA BAY GUP YAX KLY UMP, W M K YIM, GYA YDN DYI BGU BNJ GUM
"Note th a t only those levels relevant to the languages in this stu d y are shown. Thus, for ex ample, although the Kanyara-M antharta-Y uulngu subgroup contains a M an th arta sub-subgroup, no M antharta languages are used in this study, and the M antharta branch is therefore not included in this list.
Kulinic Group W EM , W OI
North Darling Group BAA
Karnic Group PIT
C entral Karnic subgroup
W estern K arnic sub-subgroup DIY
These subgroups are roughly based on O 'G rady, Voegelin and Voegelin (1966), with a num ber of m odifications. T he K anyara-M antharta-Y uulngu subgroup reflects the suspected relatively close relationship betw een languages such as BAY and GUP (O ’G rady 1959): the Nyungo-Yuulngic group contains both this and the Nyungic sub groups. Blake (1988) shows th at Y.\N. which was classified as a non-Pam a-N yungan language, is actually Pam a-N yungan, and he proposes th a t it is a m em ber of the W arluwaric group. DIY and P IT are considered to be m em bers of the K arnic group, as established by .\u s tin (1990).
.Additional modifications have been proposed since 1966. For exam ple. Wurm (1972) believes th a t GY.A. YIM. DYI and YDN are not Pam ic languages. He posits a Yalanjic group, containing GYA and YIM. a Dyirbalic group and a Yidinyic group. He also classifies UM P as a m em ber of a N o rtheastern-P am a subgroup, not of the M iddle-Pam a group. A lthough the Pamic languages have been well-studied in relation to other .Australian languages, the relationships among them have not been defini tively established. I therefore follow the subgrouping given in O ’Grady, Voegelin and Voegelin (1966) and O ’Grady. Wurm and Hale (1966), which is largely based on the work of Hale (see also Hale 1976b, 1976c, 1976d).
Work in progress by McConvell and Laughren (1996) and Laughren and Mc Convell (1996) on th e Nyungic languages has resulted in a classification differing in m any respects from th a t shown above. T he K anyara and M an th arta sub-subgroups are included as m em bers of the Nyungic group (as they were in the original classifi
cation), and their possible relationship w ith th e Yuulngic languages is not discussed. In addition, this work has suggested a num ber of relationships am ong th e Nyungic languages which would add stru ctu re to a relatively flat family tree. However, as th e authors themselves consider th eir classification to be highly tentativ e. I will continue to assum e the subgroupings shown above.
It is fully expected th a t future work will indicate that further changes to the classification of Pam a-N yungan languages are necessary. The im pact of such changes on this work would not be huge, since its p rim ary purpose is to stu d y th e developm ent of initial velars, not to establish subgroupings. . \ re-evaluation of th e proto-languages for which reconstructions are given would of course be called for.
C h a p te r 2
M e th o d o lo g y
Many interesting issues regarding the com parative m eth o d have been raised in recent years, including th e regularity of sound change and th e applicability of the co m p ara tive m eth o d to A ustralian languages. In this chapter, I will discuss several asp ects of m ethodology. F irst. I will review the traditional version of th e com parative m eth o d , and discuss some of th e problems encountered in its application. 1 will th en briefly sum m arize several views on sound change which differ in various regards from the N eogram m arian one. In Section 2.3. 1 com pare “b o tto m -u p ” and "top-down" ap proaches to com parison and describe some of the advantages and disadvantages of the la tte r direction. Finally. 1 detail the m ethod used in this study. U nfortunately, m any interesting issues relating to the com parative m eth o d and .Australian languages are beyond the scope of this chapter. For discussions on the applicability of the m ethod to unw ritten languages, see Haas (1969) and C am pbell (1996). For a review of argum ents specifically concerned with A ustralian languages, the reader is referred to Chen ( 1992).
2.1
R e v ie w o f th e C om parative M e th o d
The com parative m ethod has long been considered a very powerful and useful tool in the study of th e history of languages. Although the com parative m ethod is of ten equated w ith the Neogram m arian hypothesis, in G rim m 's tim e no distinction was made between regular and irregular sound correspondences. Toward the end of the nineteenth century, th e Neogrammarians put forward the hypothesis th a t sound change is regular: th a t is. all exceptions can be explained as cases of analogical change or borrowing (Crowley 1992:225-228). This assum ption has been th e foundation of th e com parative m ethod ever since. However, it does not always hold, as discussed below.
The m ethod is presented in detail in .\n ttila (1972). Goyvaerts (1975), Bynon (1977). Crowley (1992). McMahon (1994). Fox (1995). and m any o th er historical linguistics textbooks. In this section. I briefly outline the m ethod as it is traditionally taught, and th en discuss some problems which arise when it is actu ally applied. Crowley (1992:110-111. 228) gives the following sum m ary of the steps involved in comparing two or more languages:
1. Sort out those forms which appear to be cognate and ignore th e non-cognate forms.
2. W rite out th e full set of correspondences between the languages you are look ing at (including correspondences where the sounds are identical all the way through)
----3. Separate out those correspondences which are system atic from those which are isolated (i.e. which occur in only one or two words) and ignore th e isolated correspondences.
4. G roup together all correspondences th a t have reflexes th a t are phonetically similar.
5. Look for evidence of com plem entary and contrastive distribution between these suspicious pairs of correspondences.
6. For each correspondence set th at is not in com plem entary distribution with another correspondence set. assume th a t it goes back to a separate original phoneme.
7. Make an estim ation about the original form of th e phoneme using the following criteria:
(a) The proposed original phoneme m ust be plausible . . . .
(b) The sound th at has the widest distribution in the daughter languages is most likely to be the original phoneme.
(c) .\ sound corresponding to a gap in the reconstructed phoneme inventory of the proto-language is likely to be a possible reconstruction for one of the correspondence sets.
(d) .A. sound th a t does not occur in any of the daughter languages should not be reconstructed unless there are very good reasons for doing so.
8. For each group of correspondence sets th a t are in com plem entary distrib u tio n , assum e th at they all go back to a single p ro to -p h o n em
e----T h e first step is not always as straightforw ard as it appears, for a num ber of reasons. W hen beginning a com parative study, we cannot know what types of correspondences to expect, and this makes it sometimes difficult to judge p utative cognates. If we are com paring a large num ber of languages, or if the languages have extensive dictio naries, the num ber of forms exhibiting potentially plausible sound correspondences
can be daunting. It is therefore helpful to have some idea of th e types of sound changes th a t occur in th e fam ily or group, because changes w hich are com m on in one set of languages m ay not necessarily be so in another. For exam ple, in both the G erm anic and .\rm en ian branches of Indo-E uropean certain stops shift to fricatives; in G erm anic, plain voiceless stops change to th e corresponding fricatives as part of G rim m ’s Law (Baldi 1983:130). and in A rm enian, proto-Indo-E uropean *p becomes
h (Baldi 1983:80). Sim ilar changes are extrem ely rare in Pam a-N yungan languages,
and in fact most m em bers of th e family do not have fricatives. Som e Pam a-N yungan languages, however, drop initial consonants to an extent not seen in Indo- European. By knowing which changes are common, we can sta rt our search w ith th e m ost likely forms.
In addition, it can be very difficult to distinguish borrowings from tru e cognates, especially in the early stages of a study or when working w ith languages w ithout a w ritten history. For this reason, m any studies begin with a com parison of basic vocabulary items, such as term s for body p a rts, kinship relations, hum an actions and pronouns. It is felt th a t these types of words are less likely to be borrowed and are thus more useful for establishing sound correspondences. T his generalization, however, is not necessarily tru e of all language families. .Australian languages are well-known for banning th e use of words th a t are phonologically sim ilar to th e nam e of a person who has died, no m a tter what those words may be. For exam ple, the first person singular pronoun was replaced tw ice in ten years in th e W estern Desert language due to death taboos (Dixon 1980:29). Replacem ents a re often borrowed from neighbouring languages.
Both sound and m eaning m ust be considered in th e first step of th e m ethod, but phonological plausibility is th e more im p o rtan t aspect since m eaning can change in ways which are not obvious. This is especially tru e if the languages u nder stu d y are part of a culture different from our own. We often encounter th e problem of judg
