Chifundi and Vumba: Two Heart Attack Patients Who Survived
Chifundi and Vumba: Two Heart Attack Patients Who Survived
Chifundi and Vumba: Two Heart Attack Patients Who Survived
DerekNurseandMartinWalsh
1 Introduction
Thetargetsof thisstudyareChifundi andVumba, twosmallanddwindling Swahilidialect
communities, spokenon thecoastof southern Kenya,andof southern Kenya/northern Tanzania,
respectively. Thepurposeis to investigate (a)certaininnovations, mainlyphonological, in
ChifundiandVumba,innovations whichgivethema distinctly non-Swahili appearance and(b)
modelsthatmightexplainthediachronic appearance of lheseinnovations. We suggestthatthese
innovations derivefromhistorical interference fromDigo,a neighboring language.
ln all,four(Bantu) languages/dialects areinvolved: Chifundi(Ch),Vumba(Vu),Digo(Di),
andSegeju(Se).Ch andVu areSwahili(Sw)dialects.Di is a southern Mijikenda (MK)dialect.
Sw andMK(thusthe mainthreetargets)aretypologically similarandcloselyrelated.Se (of
Tanzania, seefollowing) is todaya formof Di,bul mostlikelyresulted froma language shiftinto
Di by a community oncespeaking Daisu,a member of theCentral KenyaBantugroup(= Kikuyu,
Kamba,Meru,etc).TherewerestilloldSegejuonthecoastin 1935whostillspokeDaisu
(Dammann 193617:231-3). Ch,Vu,andDieachhavesomeinternalvariation.
Ch is spokenin a stringof Kenyacoastalvillages between theMkurumiji andtheRamisi
Rivers, at twovillages (Mkwiro, ya
Nyuma Maji)onWasinilsland,andat Aleni,Yungi,andNgoa
justto thewestof theShimoni Peninsula. Vu is spokenat Wasinion Wasinilsland,in most
villages on theShimoni Peninsula, andin a stringof coastalvillages between approximately
Majoreni (Kenya) andtheBomaPeninsula (Tanzania). Wearenotclearexactly whereSe is
spokenbutmostlikelyin thegeneral hinterland between theborderandTangaTownin Tanzania.
Di is spokenbetween justsouthof Mombasa (Kenya) andTanga,mostlyin thehinterland, butat
somepointsDi settlements comedownto thecoast.
The1979Kenyacensusregistered 1519"Swahili/Shirazi" in KwaleDistrict, thatis,
1519peoplechoseto identify themselves to thecensustakersas "Swahili" or "shirazi"
on the
Kenyacoastbetween Mombasa andtheborderwithTanzania. Since"Swahili/Shirazi" in this
contextis moreor lesssynonymous with"Chifundi" plus"Vumba", therewereca.1519people
wholabelled themselves as Vumbaor Chifundi in Kenya.Giventherateof increase of theEast
Africanpbpulation, andthatthereareseveral "Vumba" villageson thenorthern Tanzanian @ast,
thetotaf"Chifundi/Vumba" population is probably between2,OOO and3,000. 120,024 people
registered as "Digo"in Kenya(101,336) andTanzania (18,688). Segejunumbers arenot
clear,as wewillseebelow.
Theliberaluseof inverted commas in thepreceding paragraph is meantto express doubt
aboutthe interpretation of thesefigures.Inthisareathereis oftena discrepancy betweenwhat
peoplespeakandhowtheyidentify themselves ethnically. Forexample, nearly2O,OOO people
identified themselves as "Segeju" in thelastTanzanian census (1969) thatstillreferredto
ethnicity, plusseveralthousand in Kenya.Butin Kenyanoobserver in thelastdecades hasbeen
ableto identifyanyonewhospeaksSe:in Kenyaall "Segeju" speakonlySw and/orDi - in Kenya
thereareno Se-speakers left. InTanzania in themiddle1970s,NurseandPhilippson wereable
to collectSe datathatindicated thatat leastsome "segeju" spokea formof Di different from
formsof Didescribed elsewhere, so in Tanzania Se mightbedescribed as a Didialect.Similarly,
whenoneof Walsh's co-workers in Kenyainterviewed 50 Vu-speaking womenin Shimoni, they
all identified themselves as "Digo".Hencethefigurescitedabovedo notreflectaccurately
language affiliation,
of whichwe in facthavenodetailedbreakdown. Ourbestguesstimate would
bethenumber of Vu (andpossibly Ch)speakers oughtto be slightly higherthantheethnic
identification suggests, andthattheDifigurewouldbecorrespondingly lower.Although we have
nowayof knowinghowmanypeoplespeakSe in northern Tanzania, thefactthatthe informant
L
whosupplied thedatalo NurseandPhilippson
wasyoungwouldsuggest thatthereis stillan
activeSe speaking
community, possibly
severalhundred
or thousand
strong.
2 Historicalbackground
Theconventional history
of theareaunderconsideration is asfollows3.
A lineof Swcommunities hasbeenin situforovera millenium alongtheEastAfrican
littoral.Stretching
overa thousand milesfromnorthto south,it waskeptintactby strong
economic, cullural,
religious,
andfamilyties.Thearchaeological record,supportedby a fairly
detailed oraltradition,
indicatesthatVucommunities (Vumba Khuu,seemap)haveexistedin
situsinceat leastca AD 1400.Whilethearchaeological evidence forChcommunities is harder
to interpret,number of considerations makeit likelythatCh settlements areequallyold. The
difficulty
of distinguishing
VufromChsettlements at thatremove, andthedispersed natureof
thesettlements plusthe relativeabsence of stone/coral buildingsin thearea(T.Wilsonp.
comm)makeit difficultto estimate thesizeof theseearlypopulations. Buttheverynatureof
thesettlements andtheabsence of earlyreference to themmakesit likelythattheywerenot
large,withpopulations probably hardlylargerthantoday's.At a pointunclearbutmost
probably duringthesixteenth century, theDi moveddown.from thenorthandrapidly occupied
..,
)
thehinterland.Finally,
duringthelatesixteenth,andearlyseventeenth, theSe (or
century,
Daisu)arrivedfromthenorthandsettledin the hinterland.
Intheearlydecades of theseventeenth
century,theVucommunity expandedandwiththeaid
of DiandSe,conquered theCh,capturingmanyof theirmainvillages,
andreducingthemto a
3.1 Segeju
We useSe hereto refernotto Daisu,butto theformof Di apparently
stillusedin Tanzania,
and possibly
formerly
on theKenyacoast.Ofthefourtongues underexamination,it is theone
forwhichwe haveleastdata. Thisdataderivesfroma wordlistof ca 800itemsfilledout in the
1970sby a TanzanianSefromMongavillage, followed bya shortinterview
focussing on details
of phonology
andmorphology. Themainfeatures of thisdataare:
- lexically, phonologically, andmorphologically, thisSe mostresembles Di(Nurse1982),
specifically southern Di (seebelow).
- whilethemajority of lexisis sharedwithDi,a veryfewSe lexicalitemsaresharedwith
Daisualone,andthephonology and/orgeographical distribution of theseitemssuggests originin
Daisu.ln slightly morecases,itemsaresharedby Se,Di,andDaisu4.
- of the800iexicalitemsat least150areborrowed fromSw. In a modified versionof
Swadesh's 100-word listat least13%is borrowed fromSw. As we seebelow,thephonology of
Vu andCh is quitedistinctive fromthatof therestof Sw:on thisbasis,a singleitemin 800is
borrowed fromVuor Ch. Thatis,VuandCharenotthesourceof Sw loansin Se. Themajority of
loansfromSw arefromtheZanzibar dialectUngujaor fromStSw.
- theSe informant wasmorecavalier in writing downphonologicaldetails thanotherMK
informants completing thesamelist:thus,affricates weresometimes rendered as theequivalent
fricatives, fll and[r] weresometimes usedinterchangably, thevoicedbilabialfricativewas
rendered ditferentially, etc. 15yearslaterit is hardto knowwhether thiswastheresultof the
informant's incomplete controlofthelanguage, or of hislackof experience in writingit down,
or whether it mirrored thestateof Seat thattime.At anyrate,it seemsto suggest thatthe
informant waslessawareof normsthanotherMKinformants. lt couldbe notedherethata Di
linguist at theUniversity of DaresSalaam oncereferred disparagingly to Se as a "garbled" form
of Di.
- theuseof ill and[4justmentioned is of particular interest. AllSwahili dialects are
"f-fanguages", thalis,theregular reflexof Proto-Bantu (PB).l (Guthrie's ./dl)in inherited
itemsislll (orzero).Withtheexception of Di andSe,lhe regular reflexes of this.l in
inherited itemsin all MKdialectswereoriginally [r] (now/r/)in thecontextof frontvowels,
otherwise [] (now /l/):statistically
lll predominates. In Se andDi,especially in southernDi
(thetypeof Di spokenwherethereareor wereadjacent Se communities), thedistribution of /l/
andlrlis opaque, notfollowing anyobvious or regular pallern,butwitha preponderance of
lrl. ln Daisuanditsrelatives in Central Kenya, theregular reflexof PB*l in inherited words
is/r/. Thisbehaviour of Seandsouthern formsof Diwouldseemto suggest a transfer of
articulatory or allophonic habitsfromDaisu,as nootherlanguage in thewholeareatreats*l in
thisway. Whether thesehabitsweretransferred directly intosouthern Di,or initiallyintoSe,
andlaterintosouthern Di,is notclear.
3.2 Digo
At alllinguistic
levelstheMKdialects, includingDi,arerelatively
homogeneous.Di has
undergone certainminorinnovations notsharedby thenorthernMK(Nurse1982).As mostof
thesearenotattributable to Sw influence
we do notdiscussthemhere.Sw-inducedfeaturesare:
- lexicostatistically,
Di is themostdeviant of theMKdialects.Thelexicalinnovations
in Di
aremainlytheresultof bonowing, froma varietyof sources,
mostobviously fromSw. ln the
modifiedversionof Swadesh's 1OO-wordlist,roughlythesame% derivesfromSw as in Se. ln
thetotalof ca 1600Di itemsavailable to us severalhundredareborrowed fromSw. As for Se,
mostof thesearevisiblyfromUngujaor StSw.A veryfew(5 ?l arefromVu or Ch. ThusStSw
-tetema'shiver, tremble'would regularlycorrespond to -(h)e(h)ema in Di but-rerema in Vu
andCh:in thiscasetheactualDiformis -rerema, a loanfromVu/Ch.(. CanMWprovide details
# of specificvocabulary
- an interesting
fields,e.g.farming? localflora,fauna.)
phonologicalfeature canbe observed in Di. Di speakers havean abilityto
manipulate thephonological codebetween DiandSw. Numerous itemsin Diwererecorded in two
forms,onea Swshape,theotherthecorresponding Dishape: whenaskedaboutthis,speakers
seemedto thinkit unimportant, andwereoftenunableto identifytheSwformas such. Some
wordsknownto be recentloansin Swalsoappeared in Di,butin thecorresponding Di shape,that
is,Dispeakers areeasilyableto transmute Sw material phonologically intoDi. Thisis
presumably theresultof widespread andlongstanding bilingualism in Di andSw.
- whereDi ditfersin detailsof nominal or verbalmorphology fromthe restof MK,mostof
thebasicdifferences do notappearto resultfromborrowing fromSw. A fewdetailsdo originate
in Sw:thus,thereplacement of theMK-kala'be(copula, auxiliary)'bygeneralSw-wa,andthe
lossof 3sgverbalsubject w- (stillpresent
in thelatenineteenth century).
- therearecertainsmalllexicaland phonologicaldifferences between northernDi andthe
southern Di spokenroughly fromtheChareain KenyadownintoTanzania. Whilesomeof these
cannotbeattributed to external influence,a fewcan.Thusa fewlexicalitemspresent in
southern, butnotnorthern, DiaresharedwithSe (andoftenDaisu), sowereprobably borrowed
fromSe. As pointed outin 3.1,a higherincidence of/r/ in southern Di pointsto a Se and/or
Daisusource.
3.3 Chifundi
ChandVusharea number of innovations whichdistinguish themfromtherestof Swand
makethemmorelikeMl(Di (section 3.5).Inthissection we lookonlyat Ch-specific features,
andin 3.4,at Vu-specific features.
TheSwcoastal dialectscanbedivided intoNorthern andSouthern. Ch is themostsoutherly
of theNorthern Dialects,Vuthemostnortherly of theSouthern Dialects.Justbecause thetwo
communities havesatat theNorthern-Southern divideforcenturies, bothhavenaturally
absorbed features fromtheotherset. Chis uniquein havinga number of specific
features that
canbe attributed neitherto Northern or Southern Dialects,norto Di or Se. As thesearehardly
likelyto havedeveloped in thelastfourcenturies, whenCh hasbeenhardpressed by theother
communities in thearea,thisis a reasonforthinking thatChhasa longerindependent history
thanthearchaeology wouldsuggest (see1.0).Because of intrusionbyVu,Di,(andSe ?) in the
lastfourcenturies, theChcommunity is todaysplitgeographically intothree:Funzi5, spoken
between theRamisiandMkurumiji Rivers;what couldbecalledWasini, spokenin twovillages
on eastern Wasinilsland;andreportedly in a fewvillages justto thewestof theShimoni
Peninsul{.As we havevirtually nodataforthelatter,it is ignored.FortheFunziandWasini
we havea certain timeperspective,,since Lambert (1958)is basedondatacollected in the
1920s, whereas Walshhasworkedin theareain thelasttwoyears.Walshreports theWasini
Chiclaimto haveno DiorSesettled amongthem- whereas theFunzivillages aresurrounded by
Di (andformerly Se ?) settlements. Relevant features specificto Ch,or witha distributionthat
reportedly differsfromWasinitoFunzi,are:
of velarsbeforefrontvowel(cf.cifyu'knife',masicini
- palatalization 'poor(< Arabic)',
mce'woman', -fyajia'sweep',mjeni'guest'with
otherSwkisu,maskini, mke,-fagia,mgeni).
Thispalatalizationis a regular
featureof MKdialects(except themostnortherly MK)including
DiandSe. lt alsooccursinJomvu,a Swdialectnorthof Mombasa, alsoadjacent to theMKarea.
Lambert saysit is moreregular in WasinithanFunzi.
- Lambert (1958:16)describes thefouranterior of Ch (f,v, s, z) as being
fricatives
b
affricated.In generalMKhasmoresegments thatareclearlyaffricated
thanSw:so MKts = Sw
MK dz = 5e7 9j6. However, the fouianterior
fricatives
of Sw and MK correspondregularly
ry,
(Swf, v, s, z = MKf, v, s, z),so it is unclearhowtheseChaffricated soundsareto be explained.
Walshhasobserved thattodayonlyFunzihasthesesounds,whereasLambertimpliesthatin the
1920stheywerepresent throughout Ch.
- Lambertreportsthattheformof the3plin Funzicanbe wa-or a-,whereasWasini,Vu,
Chifundi Vumba
of *p:
Reflexes
t{1 79% 65%
lpl 16% 27%
fl, pl
mixed 5% 8/"
of *t:
Reflexes
l4 83o/o 71o/o
Irl 10% 26/"
mixed[r, t] 7% 3%
i. Ch andVuarenotSwdialects andneverwere.Inotherwords,whentheinitialearlysplit
occurred betweenSw (alsoElwanaandMwani:theconservative members) andMK (alsoPokomo
andComorian:the leniting,innovaling,members), ChandVuwentalongwiththeleniting group.
Theobjection to thisscenariois that,apartfromthisfeatureandthefewothersmentioned in
3.3and3.5,basically everyothercharacteristic of ChandVu is Swahili-like. So Ch andVu
wouldhavehadto undergomassive borrowing at everylinguisticlevel,whichis possiblebut
implausible. lt is alsonotsupported by non-linguistic
factors,suchasoraltradition: MKD|
traditionsdo notclaimthattheChandVuwereonceMK,nordo thecoreChor Vutraditions.
ii. ChandVudeveloped together with,or wereheavily influencedby,Comorian beforethe
Comorians leftthemainland coastin thesecondhalfof thefirstmillenium AD,a possibility
advanced by Nurse(1982b, 1984/5). Thecentral basisforthisclaimis thatfrom*p and*t,
Comorian developed /^ll andlrl,iustas ChandVu. A secondary supportcomesfromComorian
traditionswhichclaimthattheycamefrom"Mrima": "Mrima" in Swrefersto thenorthern
Tanzanian coastandmightbe heldto include theChandVuareas- but"Mrima" in Comorian
refersto theEastAfrican coastingeneral.Theobjection to thisscenariois that,whileit is true
thereis considerable
3
evidencethatComorianandSw in general(soincluding Ch andVu),have
hadlongcontactoverthe lastmilleniumanddo sharecertainlinguistic innovations,
thereis no
evidence, otherthantheiridentical of *p and*t,of specificsharedsimilarities
treatment or
innovationsbetweenComorian andCh/Vu.
Southern
Di showslimited Daisu,possibly
lracesof materialfrom mediated
throughSe.
lf we lookat thepreceding
fromthepointof viewof donors,we see:
Chifundi.Noevidence thatChhascontributed
material
lo anyof theothercommunities,
otherthanminimal
lexicaltracesin Di,whichhowever
mighthavecomefromChor Vu.
4.0 Summaryandconclusions
Thisconference is explicitlyaboutcasesof language death.Thereareclearlycasesof death
andnear-death on thesouthern Kenyaandnorthern Tanzanian coast.CoastalDaisu(notDaisuat
Bwiti)lingeredneardeathfromthelate19thcenturyandfinallydisappeared on thecoastat
somepointtowardthe middleof the20thcentury, withtheunannounced deathof its lastspeaker.
Tanzanian Se doesnotappearto be in a healthystatetoday:if KenyanSe everexisted,it
presumably alsodiedrelatively recently. VuandCharesmall,insignificant,anddwindling
communities, theireconomic andpolitical rolediminishedby external
eventssincethe19th
century,theirdialectsrendered irrelevantin thepresentcenturyby thespreadof StSw.What
Di startedin Vu andCh,thepressures of the20thcenturywillpresumably accomplish. We are
howevernotreallyinterested in therelativemoribundness of Se andDaisu,but ratherin some
of theprocesses involved in thesicklycondition of VuandCh. ChandVu present a unique
laboralory,as theyseemto havebeendealta massive blowby Di,andyetto havesurvived.
4.1 Models
Ourfirstaim(section 1)wasto investigate theappearance of certainnon-inherited
innovations, particularly lril andlrl,in ChandVu (section 3.0).Oursecondwasto explore
modelsthatmightexplain theseinnovations (thissection).ln whatfollowswe leanheavily on
Thomason andKaufman1988,whopresenta numberof concreteproposals abouttypesof
language change,andaboutthelinguistic traceseachmightbe expected to leave.
Theysuggest fourbasicmodels of language change.Thefirstis thegeneticmodel,whereby
changeis largely inlernal, language is handed onfromgeneration to generation,andresultsin
thekindof relationship thatcanbe recovered by theComparative Method.Wetakefor granted
thatlargepartsof Di,Ch,andVucanbeexplained bythismodel,andwe arenotparticularly
interested in itsapplication here.Theotherthreetypesallinvolve contact-induced change:
pidginization/creolization;borrowing; andsubstratum influence,of which language shiftis a
centralpart. Fortworeasons, we thinkpidginization andcreolization playedno significanl role
in thesituation we areinvestigating. First,thereis noevidence forthekindof sociolinguistic
situationthattypically givesriseto pidginization/creotization. Thatis,oursituation involves
nophysical displacement of populations, noremovallo some place far removed fromhomeand
ownlanguage area,andnomixingof populations withno@mmonlanguage. Second, lhekindof
linguistic
changes thatnormally resultfrompidginization areabsent: we seenoevidence of
irregulartransmission, simplification, features thatcannotbe explained fromthelanguages
involved, replacement of synthetic by analytic features,roleof universals, or possible
application of bio-settings.
Thatleavesbonowing andsubstratum influence. WefollowThomason andKaufman in
thinkingthatwhilealmostanylinguistic feature canbetheoretically betranferred fromone
language to another,it is onlypossible to predictwhatwillbe in practicetransferred by
examining factorssuchas thesociolinguistic settingandthedegreeof typological similarity
betweenthelanguages involved.Westartwiththesociolinguistic setting,andinserttypological
similaritywhereit seemsrelevanl.Wefocusmainly on Vu,Ch,andDi,lesson Se as it doesnot
appearto haveplayeda cenlralrole,andlesson ChthanVu,for although Ch hasbeenmore
affecledthanVu bycontact,we havemorefactsabouttheVumbacommunity.
Afterthearrival of theSearoundAD 1600,thefourcommunities havehadsomefour
centuries to interact:it is possible thattheDiin factarrived somewhat earlier.TheVu andCh
todayarerestricted to a narrowstretchof thecoastsome25 mileslong. Evenallowingfor
claimsthattheyearlierlivedat pointssomewhat northandsouthof today'srange,theirareawas
nevermuchlonger.Bycontrast, theDioccupymuchof theareabetween Mombasa andTanga,ca
100miles,andextending somedistance inland.As farascanbejudged,theyarelikelyto have
occupied
muchof thisareafroman earlypoint.TheDi population
todayis over
r{
(ii)(Partial)
language shift.Vu (andCh)actedas a magnet fornearbyDi (andSe)
communities. As otherSwcommunities alongthecoast,theyconstantly absorbed individualsand
maybesmallgroups fromDi,andultimately assimilated thementirely linguistically.Whileit
is notclearif manyDiwerebilingual in theearliest stagesof therelationship between Vu and
Di,it musthavebeenthecasethatat leastthoseindividuals whowereassimilated laterwere
bilingual,andthatthiswasanongoing situation. Therewerealsoprobably Di middlemen who
werebilingual. lt shouldbe remembered herethatat leasttodayit is thenon-Swahili whotend
to bebilingual,whereas theSwaremonolingual, oftenspeaking twodialects of Sw,butrarely
another language. Givenwhatis knownaboutthecommunity's feelingof superiorityaboutitself,
thissituationcanprobably be projected backward.OntheShimoniPensinsula/there are
villages whichareVu-speaking, butin whichsections of thepopulation claimtcibe notjustDi in
origin,butalsoSe andCh. Thiswouldsuggest linguisticasgimilationof Se andChalso. Economic
dgp.etdents switched to thelanguage of theireconomic masters.At leastin somevillages(e.g.
-Kyuv0, theVudo notappearto be physically
{ lr i* t ^ ' represented in today'spopulation, andarenoteven
superior numerically in thearea:thatis,language shiftwaspossible at a shortremovefromthe
targetlanguage community.
It shouldbekeptin mindthatVuonlyactedas a magnet on adjacent partsof the Di (andSe)
communities. SincetheDicommunity in generalwas muchmorenumerous, andoccupied a
largerarea,muchof it wasnotin contactwithVu on a regularbasis.
Thomason andKaufman (pp.38-39) definesubstratum interference astheresultof
"imperfect grouplearning duringa process of language shifi...in
thiskindof interference a
groupof speakers shiftingto a targetlanguage failsto learnit correctly.Theerrorsmadeby
members of theshiftinggroupin speakingthetargetlanguage thenspreadto thetargetlanguage
t3
as a wholewhentheyareimitated by originalspeakersof thatlanguage...interference
through
imperfect
learningdoesnotbeginwithvocabulary: it beginsinsteadwithsoundsandsyntaxi1,
andsometimes includes morphologyas wellbeforewordsfromtheshiftinggroup'soriginal
ENDNIOTES
'1 Consider hereChwaka, peninsular,
spokenin severalvillages onor neartheShimoni which
appearsto be a formof Ml(iDiitselfcurrently in processof shiftingto Vu. SeeMoehligvarious.
'2 Weignoreherethethorny
issueof theterm"Shirazi". Thisis anethnicor historical, nota
linguistic,
labelusedby someSwahilicommunities. Inthecontext of localhistory, it generally
referslo a non-Vu, thatis,to theCh,andmanyChspeakers identifythemselves asShirazi, and
theirlanguage asChi-shirazi. ButmanyVualsorecognize thattheywere"shirazi" originally,
andthelermis alsoin useelsewhere on lhecoast,withdifferential reference.
-3 lt shouldbe keptin mind
thatnodetailed generalhistoryexistsforthispartof thecoast,and
thatVu is theonlycommunity of whicha reasonably comprehensive history(McKay197S)has
beenwritten.. DNto checkwithD. parkinforsperling's account of Digo/segeju.
*4 AllMKdialects, notjustDi,borrowed extensively fromDaisuin thepast(Nurse1gB2).
Thefewitemsreferred to inthetexthereareseparate fromthisgenerai loansetin MK.
*5 TheformsChifundi ([gcifu]n]dil)and(Chi)funzi mustbe related, butin anopaqueway.
Sincethe[nz]shapemustbetheoriginalform, buttheonlyNorthern Dialects in whichthis[nz]
hasshifted to [n]dl aretodayspokenin northern Kenya,manymilesaway,it is unclear why
thisformappears onthesoulhern Kenyacoast.
.6 InSwtherearestop
allophones in certainenvironments.
'7 Thisstatement is notstrictlyaccurate, asonedialectof Sw onZanzibar alsohasa voiced
labialfricativefrom.p (butrelains.t). we do notdiscuss this here.
'8 Weignorethereflexes 'mp
of and*nt,whicharedifferent.
.9 A question wedo notconsider inthispaperis whether Difeatures entered Vu andCh at rhe
sametimeor sequentially.
'10 Kinship lermsandfoodplants/types arelwoareasin whichwefoundsomeaccumulation of
loans- butourdatais notcomplete. Alsothewordfor'market'and (thefour) days of theweek.
*11 As wedo nothave
comprehensive dataforsyntax, it is ignored.
.12 Wehavesome
doubtsabouttheblanket validity of thisclaim,as muchof theverb
inflectionalsystemof Unguja/StSw seemsto resultfroma transferof featuresfromnorthern
Swwithoutanytracesof thephonological features thatoughtto precede it (Nurse19gg).
-13 lt mightbe askedwhy
otherpartsof thephonemic inventory werenotaffected.
Vu,ChandDiall had5 shortsurfacevowels,so theywouldnothaveofferedanyproblem.
Largepartsof theconsonant inventories of thethreelanguages, ascarriedin inherited
items,wereidentical, andcorresponded in cognates. Thisstatement deliberately excludes loan
words,whichmeanswordsfromArabicforsw, andwordsfromse for D|/MK.
Smallpartsof theconsonant inventories weredifferent. Thiswouldhavebeentruemost
obviously for/l/ ([] andzeroin sw, [],r] in Di,already discussed), andforMK,Did, rh,
ts,dzl= sw, Vu,ch /p,t, gc,gj/. However sw /gc,gj/wouldnothavebeentotallyaliento Di
speakers, as theydidhave[gc,gj]in inherited wordsfromothersources.
*14 Theultimate transJer to Vu andChof l^il and/rh/wouldalsohaveleadto moremarked
consonanl systems: /ph,f, b, v, il and/1,r, rh/.
.15 Theexceptions for*t areTikuu,Siu,andPate:fora discussion, seeNurse1985.For.p
theonlyexception is Makunduchi, on Zanzibari (cffootnote7):seeWhiteley1959,1960,
Lambert andchum1goz3,andNurseandHinnebusch forthcoming.
*16 Manycoastal Swahili willidentify themselves ethnically (i.e.forcensuspurposes) as
"Arab"whilespeaking littleor noArabic.
iI
REFERET{CES
Bakari,
M. 1985 TheMorphophonology of theKenyan sw Diatects.
Berlin.D. Reimer.
Dammann,E. 1936/7 'sprachproben ausdemsegedju', Zeitschrift
fuerEingeborenen-
Sprachen 27:228-233.
Lambert,H.E. 1953 'TheVumbaverb',JEASC23:14-35.
1957 Ki-Vumba, a Dialectof thesouthern Kenyacoast.Kampala. EASO.
1958 chi-chifundi,
a Dialect of thesouthern Kenyacoast.As preceding.
- /H.Chum. 196213'A vocabularyof Ki-Kae',Swahili33,1: 51-68.
Maddieson,l. 1985 Patternsof Sound.Cambridge. CUp.
McKay,W.F. 1975 A precolonial
historyof thes. Kenyacoast.Bostonu, phDthesis.
Moehlig,W.J.G. 1980 'Bantulanguages',in Heine+ Moehlig (eds.),LanguageandDialect
Allasof Kenya,vol.1. Pp.11-58.
1984/5 'Theswahilidialects of Kenyain relationto Mijikenda
andto the
Bantuidiomsof theTanaValley', SUGIA6: 253-308.
1986 'Lesparlers
bantouscotiersdu nord-est',in Guarisma,G/Moehlig
(eds.),La M6thodeDialectrom6trique
appliqu6e.... Berlin.Reimer.
- /G. Philippson/M-F.
Rombi/J.C.
Winter1980'Classification
dialectom6triquequelques de
parlers swahilis',
in Guarisma, G/S.platiel(eds),...paris.
SELAF.
Nurse, D. 1982 'Segeju andDaisu:a casestudyof a discrepancy in evidencefromoral
tradition andcomparative linguislics',
History in Africa9:175-208
1982b 'A tentative claasification
of theprimarydialects of Swahili',
SUGIA 4: 165-206.
1984/5 'A historical viewof thesouthern Dialects
of sw',suclA 6:2zs-2s2
1985 'Dentality, arealfeatures,andphonological changein Northeastern
Bantu', SAL16,3:243-279.
- /T. Hinnebusch.Forthcoming swahiliand sabaki:a LinguisticHistory.
- /T' Spear. 1985 TheSwahili: Reconstructing theHistoryandLanguage of anAfrican
Society, 800-1500. Philadelphia.
U.of Pennsylvania press.
Sperling, D. 19?? (history, lslamisation,
S. Kenyacoast).U of London, PhDthesis.
Thomason, S.G./T.
Kaufman1988Language Contact, Creolization,
andGeneticLinguistics.
Berkeley. UCPress.
Whiteley, W.H. 1959 'Anintroduction to theruraldialectsof Tanzibar, Part1',Swahili
3 0 :4 1 - 69.
1960 'dilto,Part2',
St&ahjli31: 200-218.
Wilson, T.H. 1982 'Spatial analysis andsettlements on theEastAfricancoast',in Allen,
J de Vere,andT.H.Wilson(eds.),FromZinjlo Zanzibar.paideuma
28:201-220.