TwoMissingPiecesinaNyulnyulanJigsawPuzzle

ClaireBowern,YaleUniversity.LSA2010(Baltimore)

1 TheProblem1
Most classifications of Nyulnyulan languages [nonPamaNyungan, Northern Australia] show
the family as having two branches (Bowern 2004; Stokes 1997; Stokes & William McGregor
2003;Dixon2002).However,thelanguagesinthegeographiccenterofthefamily(betweenthe
EasternandWesternNyulnyulan)areallbutunattested,andhavenotbeenusedinprevious
reconstructionsandtypologicalclassifications.Itisthereforedifficulttojudgewhetherthetwo
primary branches are the result of actual branching, or are rather an artefact of the missing
language data; it could be that we are missing data at a crucial point in a continuum which
makesthesubgroupsappearmorediscretethantheyactuallyare.Becauseoftheemphasison
Diffusionist explanations of language change in Australian linguistics, a continuum has been
assumedforthefamily(e.g.Dixon2002,66667).Irecentlydiscoveredhithertounutilizeddata
forbothmissinglanguagesandreportonresultsofreconstructionswhichmakeuseofit.
2 TheMissingLanguages:NimanburruandNgumbarl
2.1 Ngumbarl[noISOcode]
This language was thought to be unattested. The materials in the Bates ms collection (NLA
ms56526/2E.1)arefromBillingee,whoidentifiedasJukunandNgumbarl(Coyne2005).Bates
recordedmaterialforbothJukunandNgumbarl,itappears,butpreviouslyithadbeenthought
that the data were only for Jukun (Stokes & McGregor 2003:32). The materials comprise a
wordlistandafewsentencesintypescript.Thereareabout800wordsintotal,includingmany
flora,faunaandenvironmenttermsandbodyparts.Thematerialsareinaratherinconsistent
Englishspellingbased orthography and interpretation of the full collection is still in progress.
Thetranslationsarealsosomewhatunreliable(e.g.Areyouhuntingkangaroo?istranslated
by<jooainjapindana>juwainjabindanayouregoingtothepindan(scrub)).
2.2 Nimanburru[nmp]
ThereissomeNimanburrumaterialinNekesandWorms(2006)(reprintedwithadditionsfrom
1953);thisincludedsomeverbparadigmsandothergrammaticalinformation.However,Nekes
and Wormss materials are heavily standardized towards Nyulnyul; their Bardi materials, for
example,includenumerouswordsthatareclearlyNyulnyulandnotBardi(e.g.becausetheydo
notshowBardisoundchanges,becausetheyshowNyulnyulsoundchanges,orbecausethey
refertoculturalsystemswhichwerenotpresentinBardicountry).RecordingsofNimanburru
hadalsobeenmadebyAnthonyR.Peileinc.1965,howevertheyhadnotbeenauditionedor
transcribed.ThePeilecollectionhadnumerousmetadataerrors(seeBowern(2010a)).Nothing
isknownofthespeakeronthetape.NoraKerrdidsomebriefworkwithNimanburruspeakers
Djabalauruandherbrotherinthe1960s.ThiswordlistisonrestrictedaccessattheAustralian
Institute of Aboriginal Studies and so has not been used in previous historical work.
NimanburruformsquotedherearefrommytranscriptionsofPeilesrecordings,butalsofrom
NekesandWorms(1953)andKerrsmaterials.

1

ThisresearchissupportedbyNSFgrantsBCS910936,BCS902114andBCS844550,awardedtoClaireBowern
andYaleUniversity.

1
3 Comparisonoffeatures
InordertodiscoverwhereNimanburruandNgumbarlshouldbeclassifiedwithinthe
Nyulnyulanfamily,Icomparedaspectsofphonology,lexicon,casemorphology,verb
morphology,andsyntax.
3.1 Phonology
Nimanburruisveryconservativeinmostrespects,suchaspreservingProtoNyulnyulanvowel
length(e.g.*baabachild>baaba(Bardibaawa,Nyikinababa,etc;*lagu:rruegg).Thereis
some evidence that *k > / _# (e.g. PWN *bardaka tree > barda). It is difficult to infer
muchaboutthephonologyofNgumbarlbecauseofthenatureofthetranscriptionsystem.(I
have phonologized forms in part based on my knowledge of the other languages; therefore
inferencesaboutphonologyhereruntheriskofbeingcircular.)Initialsometimeswrittenask
but I assume this is not a sound change but a result of Bates difficult with nonEnglish
phonotactics.Ireconstructatentativesoundchangeofitoawordfinally(e.g.*yakiwhat>
<yanga>yaka).
3.2 Lexicon

100% MissingDataPoints
75% ProtoNyul'nRetentions
50% EasternRetentions
WesternRetentions
25%
SingleAttestations
0%
Loans

Nimanburru
Diagnostic items include (among many) marrir sister (not urnu), yalurr wifes mother (not
darlu) and bardaka tree (not ba(a)lu). Evidence from all available reconstructions indicates
thatNimanburrufitsclearlywithintheWesternbranchofthefamily,sharingmanytermsofnot
only body parts but also flora and fauna and kinship terms with other Western languages.
Currently there are no identifiable loans in the data (see figure 3 for comparison with other
Nyulnyulanlanguages).TheNimanburruvocabularyisoverwhelminglyWesternNyulnyulan.
Ngumbarl
ForNgumbarl,reconstructions(wheredifferentiatedbysubgroup)aremostlyEasternbutthere
are some Western forms as well. Ngumbarl has considerable numbers of single attestations
(18/204) and retentions from ProtoNyulnyulan with semantic shift (15/204). The four words
previouslyreconstructedtoProtoWesternonlyareasfollows:
o kunyul star (also moon): appears as moon in Western Nyulnyulan languages;
thereforethiscouldbearetentionfromProtoNyulnyulanwithsemanticshiftineither
NgumbarlorWesternNyulnyulan(theotherEasternlanguageshavethewordlarnfor
star,notelsewhererecordedinNyulnyulanoratthisstageidentifiableasaloan).
o mirrjil moon: This is probably cognate with a word for an esoteric ceremony name

2
 (mirrjil in Bardi). If not, then it should be counted as a single attestation. Note that
kunyulisalsogiveninthewordlistasawordformoon.
o jabulyuold(person):ThisisfoundinWesternNyulnyulanasthewordforgrayhair.
Other Eastern Nyulnyulan languages use the word yibala (PN *yiibala father) in this
meaning,andotherWesternlanguagesusetheword*nyuurl(PWNold,oldperson).
Thereforethisisprobablyafutherexampleofsemanticshift.
o marakub far (probably marakup, i.e. farABL): All Western languages have a reflex of
maraforthisword.OftheotherEasternlanguages,YawuruusesaloanfromKarajarri
(kajarri), and Nyikina and Warrwa both have otherwise untraceable words
(kunabid/diyadiyaandnyaarrirespectively).
Inalargersample,Ngumbarlshowsmoreitemswhichhavebeenreconstructedpreviouslyto
ProtoWestern Nyulnyulan. However, many of these are flora and fauna terms. The Eastern
Nyulnyulan languages have borrowed heavily from surrounding Marrngu and NgumpinYapa
(PamaNyungan) languages in this area of vocabulary (see Bowern 2007) and many items of
WesternfloraandfaunawerenotreconstructedtoProtoNyulnyulansimplybecauseoflackof
attestationintheavailablesourcesforEasternNyulnyulanatthetime.
3.3 CaseMorphology
Nyulnyulanlanguageshaveextensivemorphology,includingacasesystem.However,only
certaincases(suchastheproprietive)areinformativeforsubgrouping.Thecorecasesystem
canbereconstructedtoProtoNyulnyulananddevelopmentsarefoundinindividuallanguages,
notsubgroups.NimanburrushowstheWesterncomitativenyarr.Theergativecaseisnim,
fromProtoNyulnyulan*ni(ma)(regularergative+focalergativesuffix(W.BMcGregor2006;
W.McGregor2007;Bowern2003).
ForNgumbarl,littlecasemorphologyisfoundinthedata.Ergativeisna,whichiffrom*ni
showsthesameapparentsoundchangeoffinalItoathatisalsoseeninsomeotherforms.The
locativeiskun(areflexofProtoNyulnyuylan*kun).Noothercaseformsaregiveninthedata.
3.4 VerbMorphology
Nimanburru
AllNyulnyulanlanguageshaveverbswhichinflectforsubjectperson/numberandtense(along
withothercategories).MostoftheverbformsintheNimanburrudataarethirdpersonsingular
present.However,thereareformssuchasdarrunkarahewillgointhedata;thisshows
Westernfuture(transitive)nk(cfalsongankamalIllcookit),aswellasanalternation
betweenIanduforthe3sgprefix.ThisisalsoWestern,andpossiblyasharedinnovationwith
Bardi.ThereforeonthisbasisNimanburruisclearlyWestern.
Ngumbarl
There are very few verb forms in the data and no full paradigms. However, there are some
partialsingularparadigms.EasternNyulnyulanhasundergoneaclusterofchangesintheverb
prefix morphology (described in Bowern 2010b). The crucial changes are a merger of PN
presentandpast(realis)prefixparadigms,asfollows:
o PNsingularintransitivepast>Easternintransitivenonpast
o PNsingulartransitivepresent>Easterntransitivenonpast
o PNpluralpresent(transitiveandintransitive)>Easternpluralnonpast
Attested forms in Ngumbarl are consistent with such a merger, assuming that the forms are

3
given in the same tense: I steal: <kangalainbee> ngangalanybi; he steals: <ingalaimbee>
ingalanybi;theysteal:<yeeralanbee>yirrlanybi.Herethesingularformscontinueearlierpast
formsbutthethirdpluralcontinuesapresenttenseform(yirr,not*yingarr).
3.5 Syntax
TherearefewsyntacticfeaturesthatclearlyseparateEasternandWesternNyulnyulan
languages.Thelanguagesshowdifferencesbutthechangesareatthelevelofindividual
languages,andnotsubgroups.Othermorphosyntacticbehaviors,suchastheuseofinflecting
verbrootsversusbipartitelightverbconstructions,doshowanEastern/Westernsplit,but
cannotbeusedtoevaluatethepositionofNimanburruandNgumbarlbecausetherelevant
informationwasnotrecorded.
Nimanburru
Thereisnodistinguishingdata.DataareconsistentwithWesternNyulnyulan,includingtheuse
of both inflecting verbs and light verb structures, and inalienable and alienable possessive
structuresbyprefixandpossessivepronounrespectively.
Ngumbarl
ForNgumbarlthereisalsoverblittledata.However,manyverbsarequotedonlywiththefirst
part(thepreverb)ofthecomplexpredicate.Theuseofpreverbswithoutalightverbismuch
more common in Eastern languages than in Western ones. Many items that would be
translated with an inflecting verb in Western languages are found in the dataset here with a
preverb,ortheuninflectingverbinyjago(alsofoundinNyikina.)
o <ngaiinjakoolin>ayiinyjakulin.Illsleep(Igosleep);
o <kart>kardtwist
4 SummaryandConclusions
Nyulnyulanisnotadialectchain;thereisaclearsplitbetweenEasternandWesternNyulnyulaneven
when considering data from the middle languages. The languages do not show 'mixed' Eastern and
Westerncharacteristics.Atreeisanappropriaterepresentationofthefamily.Thisisfurtherevidence
thatAustralianlanguagesarenotoutsidethemethodsofhistoricallinguistics.
5 References
Bowern,Claire.2003.SomeusesofergativityinnorthwesternAustralia.InProceedingsfromtheAnnualMeeting
oftheChicagoLinguisticSociety,vol.39,622633.
Bowern,Claire.2004.Bardiverbmorphologyinhistoricalperspective.HarvardUniversityCambridge,
Massachusetts.
Bowern,Claire.2007.Eels,DolphinsandEchidnas.InAlanNussbaum&StevePeters(eds.),VerbaDocenti.
Massachusetts:BeechstavePress.
Bowern,Claire.2010a.TheNimanburruLanguageanditsPlacementintheNyulnyulanFamily.
Bowern,Claire.2010b.NyikinaParadigms:aCautionaryTaleinMorphologicalReconstruction.
Coyne,C.2005.'ByeandByeWhenAlltheNativesHaveGone':DaisyBatesandBillingee.In,UncommonGround:
WhiteWomeninAboriginalHistory,199216.
Dixon,R.M.W.2002.Australianlanguages:theirnatureanddevelopment.CambridgeUnivPr.
McGregor,W.2007.ErgativemarkingofintransitivesubjectsinWarrwa.AustralianJournalofLinguistics27(2).
201229.
McGregor,W.B.2006.FocalandoptionalergativemarkinginWarrwa(Kimberley,WesternAustralia).Lingua
116(4).393423.
Nekes,H.&E.AWorms.2006.Australianlanguages.(Ed.)WilliamMcGregor.Berlin,NewYork:Walterde
Gruyter.
Stokes,B.1997.ToptenNyulnyulanverbroots:furtherevidenceforlanguageclassification.InW.McGregor(ed),
StudiesinKimberleylanguagesinhonourofHowardCoate,17587.
Stokes,B.&WilliamMcGregor.2003.ClassificationandsubclassificationoftheNyulnyulanlanguages.InN.Evans
(ed),TheNonPamaNyunganLanguagesofNorthernAustralia:ComparativeStudiesoftheContinent'sMost
LinguisticallyComplexRegion,29.Canberra:PacificLinguistics.