Post on 07-Nov-2015
Ch
apter1
Relexificationincreolegenesisand
itseffectsonthedevelopmentofthecreole*
ClaireLefebvre
.Introduction
Thereisalargeconsensusofopinionintheliteraturetotheeffectthatcreole
languagesaremixed
languagesin
thatthey
derivesomeoftheirproperties
fromthoseoftheirsubstratumlanguagesandsomeoftheirpropertiesfrom
thoseofthesuperstratumlanguage(cf.Alleyne1981;H
olm
1988;etc.).Several
scholars,however,havenoticedthatthetypeofmixwefindin
creolelan-
guagesisnotrandom.Forexam
ple,Adam
(1883:47)statesthat:
Joseavancer...quelessoi-disantpatoisdelaGuyaneetdelaTrinidad
constitu-
entdes
dialectes
ngro-aryens.
Jentendspar
lquelesngresguinens,
transports
danscescolonies,ontprisau
franaissesmots,maisquayant
conservdansla
mesure
dupossible,leurphontiqueet
leurgram
maire
maternelles...U
netelleform
ationestcoupsurhybride...Lagram
mairenest
autrequelagram
mairegnraledeslanguesdelaGuine.1
SpeakingofHaitian
creole,Sylvain(1936:178)observesthat:
Noussommes
enprsence
dunfranaiscoul
dansle
moule
dela
syntaxe
africaine,ou(...)dunelanguewvocabulairefranais.2
Similarly,in
hisextensive
studyofFrench-based
creoles,Goodman
(1964)
observes,overandoveragain,thatparticularlexicalitem
sinthecreoleshave
aphonologicalrepresentationsimilar
toaFrench
word
butthatthey
share
properties
withcorrespondinglexicalitem
sin
theAfrican
substratum
lan-
guages.Onthebasisofdatadrawnfrom
Ndyuka,Huttar
(1971:684)
also
remarksthattheuseofmorphem
esborrowed
byapidginoracreolelan-
guage(...)from
aEuropeanlanguageoften
divergesfrom
theuse
ofthe
sourcemorphem
einthesourcelanguageandoften
correspondstotheuseof
thecorrespondingword
inthesubstratum
languages.Voorhoeve(1973)
Cla
ire
Lef
ebvr
e
mak
esa
sim
ilar
rem
ark
on
the
bas
isof
Sran
anan
dSa
ram
acca
ndat
a.K
oop-
man
(1986)
com
par
esa
num
ber
ofle
xica
lan
dsy
nta
ctic
pro
per
ties
inH
aiti
an
and
ina
sam
ple
ofW
estA
fric
anla
ngu
ages
(Kru
and
Kw
ala
ngu
ages
,as
wel
las
one
Man
de
and
one
Gur
langu
age)
.H
erco
ncl
usi
on
istw
ofo
ld(K
oopm
an
1986:2
46):
Fir
st,
W.
Afr
ican
langu
ages
shar
em
any
pro
per
ties
amongs
tth
emse
lves
,an
d
seco
ndly
,th
ese
pro
per
ties
whic
hin
clude
both
lexi
calan
dsy
nta
ctic
pro
per
ties
tend
also
tobe
char
acte
rist
icof
Hai
tian
.
Fin
ally
,K
eesi
ng
(1988:1
)w
rite
s:
Ihad
earl
ier
bee
nst
ruck
,w
hen
Ihad
lear
ned
Solo
mons
Pid
gin
inth
e1960s
thro
ugh
the
med
ium
of
Kw
aio,an
indig
enous
langu
age
Ial
read
ysp
oke
fluen
tly,
that
the
lear
nin
gta
skm
ainly
requir
edle
arnin
gPid
gin
equiv
alen
tsof
Kw
aio
morp
hem
es.T
he
synta
xofSo
lom
ons
Pid
gin
was
esse
nti
ally
the
sam
eas
the
synta
x
of
Kw
aio,
...
ther
ew
asa
virt
ual
morp
hem
e-by-
morp
hem
eco
rres
pondan
ce
bet
wee
nK
wai
oan
dPid
gin.
Thes
eobse
rvat
ions
sugg
estth
atcr
eole
langu
ages
are
notfo
rmed
by
anar
bitra
ry
mix
ture
of
the
pro
per
ties
of
the
langu
ages
pre
sent
atth
eti
me
they
are
bei
ng
crea
ted.T
he
pat
tern
that
seem
sto
emer
gefr
om
the
obse
rvat
ions
report
edab
ove
isth
efo
llow
ing:
while
the
phonolo
gica
lfo
rms
of
the
lexi
calen
trie
sof
acr
eole
com
efr
om
super
stra
tum
expre
ssio
ns,
the
sem
anti
can
dsy
nta
ctic
pro
per
ties
of
thes
ele
xica
len
trie
sfo
llow
the
pat
tern
ofth
esu
bst
ratu
mla
ngu
ages
.T
his
rais
es
the
ques
tion
ofw
hat
the
pro
cess
whic
hge
ner
ates
such
adiv
isio
nofpro
per
ties
is.O
nth
ebas
isof
Hai
tian
dat
ain
volv
ing
funct
ional
cate
gory
lexi
calen
trie
s,I
argu
eth
atth
em
enta
lpro
cess
of
rele
xifica
tion
acco
unts
for
this
div
isio
nof
pro
per
ties
ina
stra
ightf
orw
ard
way
(see
also
Lef
ebvr
e19
86,
1992
,19
93a,
b,
1998
ban
dre
fere
nce
sth
erei
n;Lef
ebvr
ean
dLum
sden
1989
,19
94).
By
its
very
nat
ure
,how
ever
,re
lexi
fica
tion
cannot
be
the
only
pro
cess
invo
lved
incr
eole
genes
is,ev
enin
the
case
ofa
radic
alcr
eole
such
asH
aiti
an.
As
ispoin
ted
out
inLef
ebvr
ean
dLum
sden
(1994),
rele
xifica
tion
applies
in
creo
lege
nes
isw
hen
the
spea
kers
ofth
esu
bst
ratu
mla
ngu
ages
are
targ
etin
gth
e
super
stra
tum
langu
age;
when
thes
esp
eake
rsst
art
targ
etin
gth
ere
lexi
fied
lexi
cons,
that
is,
the
earl
ycr
eole
,th
eyar
eno
longe
rusi
ng
rele
xifica
tion
to
dev
elop
the
creo
le.
Ithas
bee
npro
pose
dth
at,
when
the
spea
kers
of
the
subst
ratu
mla
ngu
ages
star
tta
rget
ing
the
rele
xified
lexi
cons,
two
oth
erpro
-
cess
espla
ya
role
inth
edev
elopm
entofth
ecr
eole
:dia
lect
leve
llin
g(c
f.Lum
s-
den
and
Lef
ebvr
e1994)
and
rean
alys
is(c
f.Lef
ebvr
e1984;
Lef
ebvr
ean
d
Rel
exifi
cati
on
inC
reole
genes
is
Lum
sden
1994).
The
dat
are
sult
ing
from
rele
xifica
tion
willbe
show
nto
feed
the
pro
cess
esofdia
lect
leve
llin
gan
dre
anal
ysis
.In
this
chap
ter,
Iillu
stra
tehow
thes
eth
ree
pro
cess
esap
ply
incr
eole
genes
ison
the
bas
isof
Hai
tian
dat
a
invo
lvin
gfu
nct
ional
cate
gory
lexi
cal
entr
ies.
Bef
ore
doin
gso
,I
pro
vide
a
defi
nit
ion
ofth
ese
thre
epro
cess
es.3
.
Rel
exifi
cati
on
The
men
talp
roce
ssofre
lexi
fica
tion
has
bee
nsh
ow
nto
pla
ya
role
inth
ege
nes
is
ofm
ixed
langu
ages
(cf.
e.g.
Muys
ken
1981
).T
his
pro
cess
has
also
bee
nar
gued
topla
ya
centr
alro
lein
creo
lege
nes
is(c
f.Lef
ebvr
e19
86,1
993a
,b;L
efeb
vre
and
Lum
sden
1989
,19
94).
Acc
ord
ing
toLef
ebvr
ean
dLum
sden
s(1
994)
form
al
defi
nit
ion
of
the
pro
cess
,re
lexi
fica
tion
builds
anew
lexi
con
inth
efo
llow
ing
way
.T
he
lexi
calen
trie
softh
ele
xico
ns
ofth
esu
bst
ratu
mla
ngu
ages
are
copie
d,
and
the
phonolo
gica
lrep
rese
nta
tionsin
thes
eco
pie
dle
xica
lentr
iesar
ere
pla
ced
wit
hphonolo
gica
lre
pre
senta
tions
der
ived
from
the
phonet
icst
rings
of
the
super
stra
tum
langu
age
or
by
null
form
s.T
he
seco
nd
step
isre
ferr
edto
as
rela
bel
ling.
The
choic
eofth
eper
tinen
tphonet
icst
ring
inth
esu
per
stra
tum
lan-
guag
eto
rela
bel
aco
pie
dle
xica
len
try
isbas
edon
thei
ruse
insp
ecifi
cse
man
tic
and
pra
gmat
icco
nte
xts
such
that
,as
isad
voca
ted
inM
uys
ken
(198
1),
the
sem
antics
ofth
esu
per
stra
tum
stri
ng
must
hav
eso
met
hin
gin
com
mon
with
the
sem
anti
csofth
esu
bst
ratu
mle
xica
len
try
that
isbei
ng
rela
bel
led.
Inth
elite
ratu
reon
creo
lege
nes
is,it
has
bee
npoin
ted
outth
atth
em
aker
s
ofa
creo
ledo
notid
enti
fyth
efu
nct
ional
cate
gory
lexi
calen
trie
s(i
.e.d
eter
min
-
ers,
com
ple
men
tize
rs,te
nse
,m
ood
and
aspec
tm
arke
rs,et
c.)
ofth
esu
per
stra
-
tum
langu
age
(cf.
Lef
ebvr
e1984;
Car
den
and
Stew
art
1988;
Mufw
ene
1991;
Lef
ebvr
ean
dLum
sden
1994;et
c.)
bec
ause
ofth
elim
ited
acce
ssth
atth
eyhav
e
toth
edat
a(c
f.e.
g.T
hom
ason
and
Kau
fman
1991).
InLef
ebvr
ean
dLum
sden
(199
4),i
tis
clai
med
that
the
funct
ional
cate
gory
lexi
cale
ntr
ies
copie
dfr
om
the
subst
ratu
mla
ngu
ages
are
rela
bel
led
on
the
bas
isofphonet
icst
rings
ofsu
per
-
stra
tum
lexic
al
cate
gori
es.
Itis
furt
her
clai
med
that
when
rele
xifier
sdo
not
find
any
appro
pri
ate
phonet
icst
ring
inth
esu
per
stra
tum
langu
age,
that
is,a
form
whic
his
both
sem
anti
cally
and
dis
trib
uti
onal
lysu
itab
le,
the
copie
d
funct
ional
cate
gory
lexi
calen
try
may
be
assi
gned
aphonolo
gica
lly
null
stri
ng,
such
that
when
this
lexi
calen
try
isuse
din
anutt
eran
ce,it
isnotpro
nounce
d.
The
form
alre
pre
senta
tion
ofth
epro
cess
ofre
lexi
fica
tion
pro
vided
inLef
ebvr
e
and
Lum
sden
(1994)
isillu
stra
ted
schem
atic
ally
in(1
).4
ClaireLefebvre
(1)
origin
al
lexical
en
try
target
lan
guage
[phonology] i
[sem
anticfeature] k
[syntacticfeature] n
[phoneticstring]
jused
inspecificsemantic
andpragm
aticcon-
texts
creo
le
[phonology] j
or[]
[sem
anticfeature] k
[syntacticfeature] n
(=(1)inLefebvreandLumsden
1994)
LefebvreandLumsdens(1994:10)proposalmakesthefollowingprediction
concerningwordorderincreoles:
Since
relexificationisthefirststep
insecondlanguageacquisition,theoriginal
aim
oftherelexifiersisto
reproduce
thephoneticstrings
ofthesuperstratum
language(...)Sincetherelexifiersintendtoreproducethephoneticstringsofthe
superstratum
language,thecreoleword
order
forlexicalcategorieswillbethe
wordorderoflexicalcategoriesinthesuperstratumlanguage(...).Ontheother
hand,sincetherelexifiersdonotidentifythesuperstratumfunctionalcategories,
thewordorderforcreolefunctionalcategorieswillbethesameasthewordorder
ofthesubstratumcategoriesthattheywererelexified
from.
Thedatapresentedinthischapterwillbeshowntosupporttheaboveclaims.
.Dialectlevelling
Theprocessofdialectlevellinghasbeenobserved
insituationswheredialects
orlanguagesareincontact(cf.e.g.Trudgill1986;Siegel,thisvol.).In
Lums-
den
andLefebvre(1994),itisproposedthatthisprocessplaysaroleinthe
developmentofcreolelanguages.Sincerelexificationisamentalprocess,itis
necessarilyan
individualactivity.Typically,situationswherecreolesarecreat-
edinvolveseveralsubstratumlanguages(cf.Whinnom1971).Thus,although
relexificationfrom
asinglesuperstratum
languageprovidestheearlycreole
communitywithacommonvocabulary(cf.Lumsden
andLefebvre1994),the
relexified
lexiconsfromdifferentsubstratumlanguageswouldnotbehomoge-
neousintheearlycreole.AsLumsden
andLefebvre(1994)proposed,someof
thesedifferencesmightbelevelled
outbytheprocessofdialectlevelling.
RelexificationinCreolegenesis
.Reanalysis
Reanalysisisaprocessthrough
whichaparticularphonologicalform
associated
withonelexicalentrycomes
tobeassociated
withanother
lexicalentry
(Lightfoot1979).Thisprocess,sometimesreferred
toasgram
maticalization(cf.
e.g.Sankoff1990;H
opperandTraugott1993),hasbeenshowntoplayarolein
casesofregularlinguisticchange.Forexam
ple,theprepositionofinEnglishhas
beenreanalyzed
asacase
marker(Chomsky1981).Likewise,accordingto
Kaynes(1981)
analysis,theFrench
form
sanddehaveadoublestatusas
prepositionsandascomplementizers.LefebvreandLumsden
(1994)
propose
that,when
speakerstargetthespeech
ofthecreolecommunity,thatis,theearly
creole,reanalysisplaysaroleinthefurtherdevelopmentofthecreole.They
claim,however,thatintheearlycreolethisprocessappliestoalexicalentry
thathasbeencreatedthrough
relexificationbutassigned
aphonologicallynull
representation(cf.(1)).AsLefebvreandLumsden
(1994:13)putit:
Intheabsenceofvisiblephonologicalsignalsforaparticularfunctionalcategory,
speakers[oftheearlycreole]useperiphrasticconstructionstoclarifyinform
ation
thatisnotbeingsignalled.Speakerswillthen
copythephonologicalform
ofthe
key
elem
entoftheperiphrasticconstructionintothelexicalentryoftheprevi-
ouslyhidden
functionalcategory.Inthisway,reanalysisprovidesan
explicitsignal
foracreolelexicalentrythatwasgenerated
byrelexificationbutdidnotacquire
aphonologicalsignalthrough
relabelling...When
reanalysisassignsthislexical
entryaphonologicalsignal,so
thatitbecomesexplicitinthecreoleexpression,
thecreolelexicalentryisvisiblymorelikethatofthesubstratumlanguage.
Inlightofthistheoreticalbackground,Iwillnowshowhowthesethree
processesareimplementedinthegenesisofHaitian
creole.Ibeginwithphase
1when
adultnativespeakers
ofdifferentsubstratum
languages
target
the
superstratumlanguageandusethementalprocessofrelexificationinorderto
createanew
language.Then
Iillustratetheprocessesthatapplyinthesecond
developmentalphaseofthecreolewhen
thespeakers
target
therelexified
lexicons,thatis,theearlycreole.
.Phas
e:
Adult
nat
ivesp
eaker
sofsu
bst
ratu
mla
ngu
ages
targ
et
thesu
per
stra
tum
langu
age
Thissectionillustrateshowrelexificationappliestofunctionalcategorylexical
entriesduringtheperiodwhen
nativespeakersofsubstratum
languagesare
targetingthesuperstratumlanguage.Caseswherethecopiedlexicalentryisre-
Cla
ire
Lef
ebvr
e
label
led
on
the
bas
isof
asu
per
stra
tum
phonet
icst
ring
and
case
sw
her
eth
e
copie
dle
xica
len
try
isas
sign
eda
phonolo
gica
lly
null
form
willbe
dis
cuss
edin
turn
.
.
Copy
and
rela
bel
on
the
bas
isof
super
stra
tum
phonet
icst
rings
..
The
[+defi
nit
e]det
erm
iner
This
sect
ion
argu
esth
atth
ele
xica
len
try
of
the
Hai
tian
det
erm
iner
has
bee
n
crea
ted
thro
ugh
the
pro
cess
of
rele
xifica
tion.T
he
dat
aan
dan
alys
isre
port
ed
on
inth
isse
ctio
nar
edra
wn
from
Lef
ebvr
e(1
994a
),bas
edon
ase
ries
ofpap
ers
on
the
vari
ous
face
tsof
this
Hai
tian
lexi
cal
entr
y(e
.g.
Lef
ebvr
e1982,
1992,
1996a,
1998a;
Lef
ebvr
ean
dM
assa
m1988).
Hai
tian
creo
lehas
apost
nom
inal
det
erm
iner
la(w
ith
the
phonolo
gica
lly
condit
ioned
allo
morp
hs
a,an
,nan
and
lan),
asillu
stra
ted
in(2
).T
he
pre
sence
ofth
isdet
erm
iner
indic
ates
that
the
info
rmat
ion
conve
yed
by
the
noun
phra
se
ispar
tof
the
shar
edknow
ledge
of
the
par
tici
pan
tsin
the
conve
rsat
ion
(cf.
Fourn
ier
1977;
Lef
ebvr
e1982;
Lef
ebvr
ean
dM
assa
m1988).
The
Hai
tian
det
erm
iner
isnot
mar
ked
for
gender
.
(2)
a.ti
mounn
nan
Hai
tian
child
det
the
child
(in
ques
tion/t
hat
we
know
of)
b.
liv
laH
aiti
an
book
det
the
book
(in
ques
tion/t
hat
we
know
of)
Inco
ntr
ast,
the
Fre
nch
det
erm
iner
appea
rsbef
ore
the
noun,as
show
nin
(3),
and
itis
spec
ified
for
gender
and
num
ber
.Le
ism
ascu
line
singu
lar,
lais
fem
i-
nin
esi
ngu
lar,
les
isplu
ral,
and
lis
aphonolo
gica
lly
condit
ioned
allo
morp
h.
(3)
a.l
enfa
nt
Fre
nch
det
child
the
child
b.
lelivr
e
det
book
the
book
c.la
table
det
table
the
table
Relex
ifica
tionin
Creole
genesis
d.
les
livres/tab
les
detbooks/tables
thebooks/tables
Inco
ntrastwithth
eHaitian
determiner,th
eFrench
determiner
does
notne-
cessarilyiden
tify
old
orknowninform
ation.Accord
ingto
Milner
(1978:23),
thedefi
nitedeterminer
iseith
eran
aphoric,
iden
tifyingan
object
that
alread
y
has
been
men
tioned
,orca
taphoric.
Inth
elatter
case,larticle
annonce
une
relative
ouung
nitifsansquaucu
nemen
tionan
trieu
reneso
itrequise.
TheHaitian
determiner
cannotap
pea
rwithnounsth
athav
eage
neric
or
massinterp
retation,butth
eFrench
determiner
must
appea
rwithsu
chnouns
(cf.M
ilner
1978:2
5).
Thesefactsareillustratedin
(4)an
d(5
),resp
ective
ly.
(4)
Pen
bnpoulasante.
Haitian
bread
goodforhea
lth
Bread
isgo
odforoneshea
lth.(=
(19)in
Lefeb
vre1994a)
(5)*(Le)
pain
estbon
pourla
sant
French
(det)bread
isgo
odfor
dethea
lth
Bread
isgo
odforoneshea
lth.(=
(24)in
Lefeb
vre1994a)
Furthermore,F
rench
has
apartitive
determiner
de+
laor
du(a
contrac
tedform
ofde+
le),
whichap
pea
rswithmassnouns,
asin
(6).
(6)
Jean
aman
gdu
pain.
French
John
au
xea
tde+
lebread
Johnatebread
.(=
(25)in
Lefeb
vre1994a)
Accord
ing
toHaa
se(1
965),
this
partitive
determiner
has
been
attested
in
French
since
thefifteenth
centu
ry.M
ilner
(1978:2
4)points
outth
eex
cept-
ional
charac
terofFrench
with
resp
ectto
this
partitive
determiner
andnotes
that
inmost
langu
ages
thedeterminer
does
notap
pea
rin
contextswhereth
e
French
partitive
determiner
isman
ifested.Haitian
followsth
epattern
ofth
e
majority
oflangu
ages,as
shownin
(7),
theHaitian
counterp
artofth
eFrench
sentence
in(6
).
(7)
Jan
man
jepen
.Haitian
Johnea
tbread
Johnatebread
.
Finally,in
Haitian
,th
ehea
dnounan
dth
edeterminer
may
beseparated
bya
relative
clau
se,as
in(8
).
ClaireLefebvre
(8)
Mounn[
ki
pati]a.
Haitian
man
op
re-proleave
det
Theman
wholeft.(=
(20)inLefebvre1994a)
Bycontrast,in
French,theheadnounandthedeterminermay
notbesep-
arated
byarelativeclause,asshownbytheungram
maticalityofthesentence
in(9).
(9)*Le
[quiest
parti]homme
French
det[who
au
xleave]man
Lit.:The[wholeft]man(=
(26)inLefebvre1994a)
Thedeterminer
inHaitian
creoleandthedeterminer
inFrench
thushave
quitedifferentsemanticandsyntacticproperties,whichindicates
that
the
properties
oftheHaitian
creoledeterminer
arenotderived
from
French.
Moreover,theFrench
determinerdoesnotappeartohavebeenthesourceof
thephonologicalrepresentationoftheHaitian
determinereither.TheFrench
determinerisoften
foundaspartofHaitian
simplenouns,asshownin(10).
(10)
Hai
tian
nou
ns
Cor
resp
ondin
gFre
nch
DPs
larivy
river