Professional Documents
Culture Documents
http://www.jstor.org/stable/4167857 .
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend access to Language
in Society.
http://www.jstor.org
IN AUSTRALIAN
LANGUAGES
Mostof thenonprefixing
languagesof Australiahaveverbalderivational
suffixes
whichderivean intransitiveverb with reflexiveor reciprocalmeaningfroma
transitiveverb.Forexample,Dyirbalhasseparatereflexive(I) andreciprocal(2)
derivationalsuffixes,as illustratedin the followingsentencestakenfromDixon
(I980:303,
(1)
bayi
433),1
yara
gunba-y:rri-nyu
baagu
barri-ggu.
bayi
yara-rrji
gunbal-gunbal -nbarri -nyu
barri-ggu.
HE-ABS man-PLURAL-ABS cut -REDUP-RECIP-PRES axe -INST
The menarecuttingeachotherwithaxes.
321
ALAN
DENCH
FUNCTIONS
OF THE NGAYARDA
COLLECTIVE
SUFFIX
KINSHIP
AND
TABLE I.
COLLECTIVE
ACTIVITY
Collective suffixforms
Conjugation
Y/0
Martuthunira
Panyjima
Yinyjiparnti
-yarri
n/a
n/a
-rnmarri
n/a
-marri
-yarri
-yarria
-lwarrib
-nyayi
-marri
-nmayi
-nmarri
-nmarri
.nyjarriC
jugationareconditioned
by thelengthof theverbstem.Dimoricmonomorphemicverbstemstake the -yarrisuffix while monomorphemic
stemsof morethantwo moratakethe -lwarri suffix.
cThe-nyjarriY-conjugation
variantin Yinyjiparnti
also appearsto be
quiterestricted.Wordick(1982:90)suggeststhatit is theformchosen
for transitiveverbswitha fmali or u vowel. His dictionaryonly lists
twoverbswhichtakethissuffix:yungku-'give'andwanyjaarri'hear,
listen'.Theverbngarrku-Y
'eat'whichwouldappearto fit Wordick's
criteriafor -nyjarritakesthe moregeneral-marriform.Wordickdescribesa furthervariantof the-nyjarriformwhichappearson twansitive
verbsof theY-conjugation
havinga finala vowel.Again,he listsonly
the one examplewangka-'speak'.
(WordickI982).2 TheKurrama
dataareincompletebutappearto mostresemble
Whilethe suffixesin the differentlanguagesappearto be cognate,
Yinyjiparnti.
it is not clear that a single set of forms shouldbe reconstructed
for protoNgayarda.3
The suffixformsin all threelanguagesare subjectto some morphophonemic
alternation.First, the L-conjugation
form is shortenedto the syllable-rril-yi
following the causativederivationalsuffix -ma-L. Second, the Martuthunira
formsmergewitha followingRelativesame-subject
inflectionalsuffix-rra.For
canproducethefollowingderivational
example,thesetwo rulesin Martuthunira
sequence:
*muthumuthuma -yam
-rra
cool
-CAUS-COLL-RELss
*muthumuthu
-ma-rri-rra
*Muthumuthu-ma-rra
muthumuthumarra
cool+CAUS+RECIP+RELss
ALAN DENCH
on intransitiveverb stems (in all examples that follow, collective verbs will be
underlined).
(3)
thawura-la -rru.
puni-marri -layi tharrwa-lu
kulhampa-ngara
-LOC-NOW
-PLURAL go -COLL-FUT enter -PURPss net
fish
The fish will all swim together into the net.
(4)
(6)
mungka-yarri-waa.
nganarna murla-a wantha-lwayarapawulu-ngara -a
-COLL-PURPs=o
lpl(exc) meat-ACC leave -HABIT child-PLURAL-ACCeat
We used to leave the kids some meat so they could have a feed.
wiruwanti yirla karlwa-marri-layi, ngartil waruul mungka-yarri -layi ngurnu
eat
-COLL -FUT thatACC
morning only get up-COLL -FUT again still
tharnta-a.
euro -ACC
In the morningwe'll all get up, and we'll still have anotherfeed of thateuro (hill kangaroo).
(7)
(9)
324
ngatha
jilya-kutha -ku
lsgNOMsee-PASTthat -DUAL-ACCchild-DUAL-ACC
thali-nmayi-jangu.
kick -COLL -REL
katama-yi-ku.
Panyjimamakes little use of the collective reading. Even the appearanceof the
suffix on intransitivestems, which require the collective reading, is relatively
infrequent.Nevertheless, the use of the label "collective" is justified, given the
extended uses of the suffix to be discussed in the next section.
To summarise so far, we have identified a verbal suffix in the Ngayarda
languages which lends a general collective reading to the verb to which it is
attached.We might best describe this suffix as a morphemederiving a new verb
lexeme which requiresa nonsingularsubject and has the added meaning that the
activity is performedtogether by the participantsdenoted by the subject NP.
Because the collective suffix may occur on intransitiveverb roots, and because a
transitiveobject may appearwith a verb bearingthe suffix, it cannotbe described
as an intransitivisingdevice like reflexive/reciprocalsuffixes in other Australian
languages.
This departurefrom an expected patternis due in part to the status of transitivity as a syntactic category in the Ngayardalanguages generally. Now, transitivity has been the subject of much attention in recent linguistic theoretical
discussion. In particular,an influentialpaper by Hopper and Thompson (I980)
has questioned the traditionalview that it is possible to make a clear division
between transitive and intransitiveconstructions. They argue that transitivity,
which is properly a feature of clauses, is affected by a number of different
properties of a predicate and its arguments. This results in a cline whereby
particularconstructionscan be considered more or less transitivein comparison
with other constructions. It follows, also, that transitivitycontrastswill assume
greateror lesser importancein different languages, depending on the degree to
which inherenttransitivityfactors are reinforcedand employed in the organization of syntax.
In most Australianlanguages transitivityis an extremely pervasive syntactic
category, largely due to the fact that these languages have an ergative type casemarkingpattern,in which transitivesubjects are markedin one way, intransitive
subjects and transitiveobjects in another(the same) way. However, unlike most
Australian languages, those of the Ngayarda group have evolved a nominative/accusative case-markingsystem (Dench I982b), a change which has resulted in a reductionin the importanceof transitivityin these languages. The loss
of a strong transitivity contrast has meant that the reciprocal suffix has no
importantpartto play in syntacticorganization,and this freedomhas allowed the
325
ALAN
DENCH
COLLECTIVE
ACTIVITY
datado
A large numberof examples appearingin the Panyjimaand Martuthunira
not conform to the general characterisationof the "collective" suffix given in
the last section. The following Panyjimaexamples show thatthe collective suffix
may appear on verbs in clauses with singular subjects and for which no clear
collective meaning is understood.
(12)
(13)
(14)
(15)
While it may be possible to read some notion of collective activity into (I2 &
13), perfonned by the subject togetherwith the object of the verb, this is clearly
not possible for (14 & I5). The following Martuthunira(I6 & 17) and Kurrama
(i8) examples illustratethe same pattem.
(16)
326
KINSHIP
AND
COLLECTIVE
ACTIVITY
KINSHIP
SYSTEMS
ALAN DENCH
Panaka
Milangka
FIGURE I:
Karimaffa
==
Purungu
two sections. For example, the agnatic kin of a person in Panakasection will be
in either Panakasection or Milangka section. Thus the Panaka/Milangkacouple
comprises one patrimoiety, the other patrimoiety being Karimarra/Purungu.
Similarly, a Panakaperson's uterine kin will be in the Panaka/Purungucouple.
The Panaka/Purunguversus Karimarra/Milangkadivision correspondsto a division into two matrimoieties.
Finally, the four sections can be grouped into two couples on the basis of
generationmembership. From Figure i we can see that in any generationonly
two sections will be represented.For example, all membersof a Panakaperson's
generation - siblings, parallel cousins, and cross cousins - will be in either
Panakaor Karimarrasections. All membersof a Panakaperson's parents'generation will be in either Milangkaor Purungusections, and similarlyfor the Panaka
person's children's generation. However, a Panaka person's grandparentsand
grandchildrenwill be in the same generation couple: Panaka/Karimarra.This
division gives two merged alternategeneration sets, the first comprising ego's
siblings, cousins, grandparents,and grandchildren,the other comprising ego's
parentsand children as well as their siblings, cousins, grandparents,and grandchildren. We can talk of people as being in the same set of merged alternate
generations or in different sets.
To turnthis discussion on its head, we see that the four-section system results
from the interactionof the two principal divisions into patrimoietiesand matrimoieties, and yields the importantdivision of kin into two alternatinggeneration
sets. In analysing complex kinterms, kinship-sensitivepronominaluse, and also
in describing institutionalisedsocial interaction, it is found that section membershipby itself is relatively unimportant.Instead, the importantcategorisations
are moieties and the alternategenerationcouples. This will be discussed further
in the section on ethnographicexplanations.
The Martuthunirakinship system involves the same importantdivisions into
moieties and generation sets as does the Panyjima system. However, where the
328
Panaka
Karimarra
Pal. yarri
Purungu
===
FIGURE 2:
Martuthunirasections.
ALAN
Purungu
Kafimarra
DENCH
= = = =
FIGURE 3:
Panaka
Palyim
Yinyjiparntisections.
In this example the use of the collective suffix reflects the fact that the subjectof
the sentence is in the same generation set as his younger brother.If a kinterm
such asfather or daughter were to be used instead, then the use of the collective
suffiixin this sentence would not be acceptable. For example:
(20)
(21)
By contrast, in sentences (i5), (i6), and (17), the collective suffix is the only
indicationthat the participantsare in the same set of alternatinggenerations.For
330
KINSHIP
TABLE 2.
AND
COLLECTIVE
Samegeneration
ACTIVITY
(inclusive)
(exclusive)
ngayu
Different generation
Dual
Plural
ngali
ngaliwa
ngaliya
nganajumarta
nganarna
nganajumartangara
nyiya kq4a-nyayi
-ku
-ku.
The linguisticcodingof the same/differentgenerationrelationshipis not restrictedto the use of the collectivesuffix in these languages.Like manyother
Australian
languages,thoseof theNgayardagrouphavea set of specialpronouns
to certaingroupsof kin. The most importantdeterminingfactorin
appropriate
the choiceof the appropriate
pronounin manyof the languagesthathavesucha
set is the same/differentgenerationrelationshipbetweenthe pronounreferents
(see, for example, Hale [I966]; Alpher[I982]). Table 2 presentsthe Martuthunirafirstpersonpronounswhichclearlyillustratethe generationsplit.
The pronounparadigmsof bothPanyjimaandYinyjiparnti
show an identical
Inaddition,secondpersonpronounsin bothYinyjipamtiandPanyorganisation.
jima havekin-determined
altemativesand, in Yinyjiparnti,
dualdemonstratives
have separatefonrs determinedby generation(for an extensiveanalysisof the
Panyjimasystemsee Dench[1982a]).
Ethnographic explanation
ALAN
DENCH
expect these large superclassesto determineinterpersonalinteractionat the individual level, where people are organised into groups, for the performanceof
ritual for example, then we might expect institutionalisedpatternsof behaviour
reflecting group divisions.
Tonkinson (1978), in discussing the social organisation of the Mardudjara
(who have a kinship system very like the Panyjimaand with whom the Panyjima
interact), makes the following general statements.
Nowhere in the WesternDesert are matrimoietiesfound, and there are neither
corporate groups nor social entities based on this kind of dual division. At
large gatherings, the Mardudjaragroup their camps into two patrimoiety
"sides," also used in the seating arrangementsfor certain men's rituals and
seen in intergroup gift exchange. But the most importantdivision is the
merged alternate generation levels, which figure prominently in many religious activities (Tonkinson 1978:57).
As for the sections, section names are used as terms for addressand referencebut
groups comprising only members of one section are not of great importancein
any ritual activity. With regard to the Ngayarda situation I know of only one,
very restricted, context in which groups are organised on the basis of membership of the one section.4
Tonkinson's general statementswith regardto Mardudjarasocial classes also
hold for the Ngayarda groups. Matrimoieties as distinct social groups are not
found although, as mentioned above, matrilinealrelationshipsare importantin
organising marriage. Patrimoietiesare in evidence in a numberof areas: traditional camping arrangementsare determinedby patrimoietalaffiliation, and access to certain increase sites is occasionally dependent on patrimoiety membership. However, as with the Mardudjara,the merged alternategenerationsets
form the most importantdivision in the organisationof ritual interactionin the
Ngayarda language area. The clearest example of this appears in the religious
activities organised aroundmale initiation, which is still practisedby the Panyjima and Yinyjiparnti. A brief outline of the activities involved in a typical
Pilbarainitiationmeeting will serve to illustratethe relationshipbetween the two
groups.5
KINSHIP
AND
COLLECTIVE
ACTIVITY
DENCH
ALAN
kungama-yi
-ma
karntinmarra-la.
tree
stand-PRES
hairbelt that
-COLL-IMP
get
Get thatbelt hangingin the tree.
-LOC
. .
is possible,whereas"otherside" is composed
interaction
tivelyunrestrained
mainly of people with whom one interactsasymmetrically(Tonkinson
I978:57-58).
Manyof the examplesinvolvingthe collectivesuffixconformto the spiritof
The use of the suffix marksthe existenceof the particular
this generalisation.
but at the sametime invokesthe contrast
relationshipbetweenthe participants
of the samegenerationset andthat
betweeninteractiontypicalof membership
of differentsets.
typicalof membership
DISCUSSION
TABLE
Verb
Subject
Object
Interpretation
reciprocal
Transitive
Transitive
Intransitive
Transitive
Transitive
Singular
Singular
Nonsingular
Nonsingular
Nonsingular
collective
kin group
yes
yes
yes
yes
yes
yes
yes
yes
no
yes
-
no
yes
yes
ngayu
kangku-yarri-iha
panaka-ngurni karimarra-wuyu-u
IsgNOM take -COLL-PAST section-BEHIND section -SIDE-ACC
marrari -mulyarra, martuthunira -a
nhuura-npa-waa.
language-ALL
language name-ACC know-INCH-PURPs=o
I took the karimarra section boy along behind the panaka boy towards the language, to
learn Martuthunira.(I taught two boys who are together in the same generation set.)
(26)
This suggeststhatwe neednotestablishseparatecollectiveandreciprocalmeanings for the suffix. Rather,a single collectivemeaningwill allow a reciprocal
in certainsyntacticcontextsandwith certainverbs.
interpretation
335
ALAN DENCH
-a.
talk
-COLL-REEts-ACC
He still can't hear us talking together (to one another).
Clearly, the kin group meaning of the suffix must be independentof the collective meaning and we must describe the suffix as polysemous between these two
meanings. On the one hand, the suffix indicates that the action is performed
together by members of a group. On the other hand, the suffix indicates that
participantsinvolved in the action denoted by the verb are in the same alternating
generation set.
Since very few sentences with a nonsingularsubject are like (27) in clearly
indicatingthe relationshipsamong participants,there is some potentialfor ambiguity in clauses with nonsingular subjects. Although such clauses are usually
given a collective reading, in some cases the suffix may serve to emphasise the
kin relationships of t'ie participants.While the collective interpretationis obviously dependent on a nonsingularsubject, it is not so easy to decide on the
availabilityof the kin group interpretation,and thus the possibility of ambiguity,
from formal features in the clause. First, as we have seen from examples (15)(17), overt specification of kin relationshipsin the form of pronounsor kinterms
need not occur in the utterance.Second, there are no syntacticrules determining
which of the availableparticipantsin the clause are linked by the suffix appearing
on the verb (see example (24)). Deciding which of a numberof participantsare
linked by the use of the verbal suffix, and whether the collective readingof the
kin group reading is intended in any particularinstance depends on close attention to other cues in discourse and in extralinguisticcontext.
The suffix rarely adds any information about kin relationships that is not
already known. In the social context in which the Martuthuniraand Panyjima
languages are spoken every person with whom one is ever likely to interactis
fitted into a complex and fluid kinshipnetworkwhich situatesevery individualin
relation to every other individual. Strangers,such as visiting Aboriginalpeople
from fartherafield, and of course the occasional linguist or anthropologist,must
be fitted into this system before any "real" interaction can take place. The
collective suffix rarely, if ever, functions to uniquely identify a participantin
discourse. Typically the individuals referredto are known to both speaker and
addressee, and their kin relationships to one another and to the speech act
participantsare also known. The suffix, then, adds no new information.Its use
336
KINSHIP
AND
COLLECTIVE
ACTIVITY
ALAN
DENCH
NOTES
*
The data were collected with the assistanceof Percy Tuckerand HerbertParker(Panyjima)and
Algy Paterson(Martuthuniraand Kurrama).I also thankDell Hymes, HaroldKoch, Bob Tonkinson,
Anna Wierzbicka, and David Wilkins for their very helpful comments and suggestions. Any errors
remain solely my responsibility.
I.
The following list expands abbreviations used in sentence examples in the paper: X (first
person), 2 (second person), ABSolutive (intransitive subject and transitive object), ACCusative,
ALLative, ASSERTedly, REFLexive, PRESent, Bro(brother), CAUSative, COLLective,
DECeased, DEFinite, dl (dual), ds (different subject), EFFector(passive actor), exc (exclusive), Fa
(father), FOREgroundingclitic, FUTure, HABITual, IDentifying clitic, IGNORantly, inc (inclusive), INCHoative, INSTrumental, LOCative, Mo (mother), NOMinative, NOW (temporal
clitic), PASSP (passive perfective), pi (plural), PN (proper nominal marker), PROPrietive,
PRIVative, PURPosive, RECIProcal, REDUPlication, RELative, SIDE (establishes an opposition
with a numberof "sides"), ss (same subject), s=o (subject coreferentialwith object), TOPic clitic.
2.
The collective suffix in Yinyjiparntiis glossed as "collective/reciprocal" (Wordick 1982:82). I
reservingthe label
have previouslydescribedthe Panyjimasuffix as "reciprocal" (Dench I98I:I20),
"collective" for a more restrictedsuffix with a somewhat similar function. This second suffix is not
relevant to the present discussion.
There is a remarkabledegree of cognacy among reflexive/reciprocal forns across the Aus3.
traliancontinent. For example, notice the similarity among the Ngayardaforms -yarri, -lwarri, and
-nmarri, and the Dyirbal forms -yirri (reflexive) and -nbarri (reciprocal).
As partof thepurntut/purntulpaceremony,menof the one sectionoperateas a marauding
4.
11913])andthe
tying" initiationritualcommonto the Ngarlumato theirnorth(Radcliffe-Brown
did initiateyoung
Mantharta
peoplesto theirsoutheast.Whileit is assertedthatthe Martuthunira
men, the detailsof the initiationprocessarenot known.
of the many
The initiationlaw practisedin the Pilbaratodayis in manyways a rationalisation
different laws originally practisedin the area;borrowing from Yinyjiparntilaw, Panyjimalaw, and
338
KINSHIP
AND
COLLECTIVE
ACTIVITY
Hopper, P. J., & Thompson, S. A. (1980). Transitivity in grammar and discourse. Language
56:25 I -99.
339