You are on page 1of 17

Introduction

Similarly, in his extensive study of French-based creoles, Goodman (1964)


observes, over and over again, that particular lexical items in the creoles have
a phonological representation similar to a French word but that they share
properties with corresponding lexical items in the African substratum languages. On the basis of data drawn from Ndyuka, Huttar (1971: 684) also
remarks that the use of morphemes borrowed by a pidgin or a creole language (. . .) from a European language often diverges from the use of the
source morpheme in the source language and often corresponds to the use of
the corresponding word in the substratum languages. Voorhoeve (1973)

Nous sommes en prsence dun franais coul dans le moule de la syntaxe


africaine, ou (. . .) dune langue w vocabulaire franais.2

Speaking of Haitian creole, Sylvain (1936:178) observes that:

Jose avancer . . . que les soi-disant patois de la Guyane et de la Trinidad constituent des dialectes ngro-aryens. Jentends par l que les ngres guinens,
transports dans ces colonies, ont pris au franais ses mots, mais quayant
conserv dans la mesure du possible, leur phontique et leur grammaire
maternelles . . . Une telle formation est coup sur hybride . . . La grammaire nest
autre que la grammaire gnrale des langues de la Guine.1

There is a large consensus of opinion in the literature to the effect that creole
languages are mixed languages in that they derive some of their properties
from those of their substratum languages and some of their properties from
those of the superstratum language (cf. Alleyne 1981; Holm 1988; etc.). Several
scholars, however, have noticed that the type of mix we find in creole languages is not random. For example, Adam (1883: 47) states that:

Claire Lefebvre

Relexification in creole genesis and


its effects on the development of the creole*

Chapter 1

These observations suggest that creole languages are not formed by an arbitrary
mixture of the properties of the languages present at the time they are being
created. The pattern that seems to emerge from the observations reported above
is the following: while the phonological forms of the lexical entries of a creole
come from superstratum expressions, the semantic and syntactic properties of
these lexical entries follow the pattern of the substratum languages. This raises
the question of what the process which generates such a division of properties
is. On the basis of Haitian data involving functional category lexical entries, I
argue that the mental process of relexification accounts for this division of
properties in a straightforward way (see also Lefebvre 1986, 1992, 1993a, b,
1998b and references therein; Lefebvre and Lumsden 1989, 1994).
By its very nature, however, relexification cannot be the only process
involved in creole genesis, even in the case of a radical creole such as Haitian.
As is pointed out in Lefebvre and Lumsden (1994), relexification applies in
creole genesis when the speakers of the substratum languages are targeting the
superstratum language; when these speakers start targeting the relexified
lexicons, that is, the early creole, they are no longer using relexification to
develop the creole. It has been proposed that, when the speakers of the
substratum languages start targeting the relexified lexicons, two other processes play a role in the development of the creole: dialect levelling (cf. Lumsden and Lefebvre 1994) and reanalysis (cf. Lefebvre 1984; Lefebvre and

I had earlier been struck, when I had learned Solomons Pidgin in the 1960s
through the medium of Kwaio, an indigenous language I already spoke fluently,
that the learning task mainly required learning Pidgin equivalents of Kwaio
morphemes. The syntax of Solomons Pidgin was essentially the same as the syntax
of Kwaio, . . . there was a virtual morpheme-by-morpheme correspondance
between Kwaio and Pidgin.

Finally, Keesing (1988:1) writes:

First, W. African languages share many properties amongst themselves, and


secondly, these properties which include both lexical and syntactic properties tend
also to be characteristic of Haitian.

makes a similar remark on the basis of Sranan and Saramaccan data. Koopman (1986) compares a number of lexical and syntactic properties in Haitian
and in a sample of West African languages (Kru and Kwa languages, as well as
one Mande and one Gur language). Her conclusion is twofold (Koopman
1986: 246):

Claire Lefebvre

The mental process of relexification has been shown to play a role in the genesis
of mixed languages (cf. e.g. Muysken 1981). This process has also been argued
to play a central role in creole genesis (cf. Lefebvre 1986, 1993a, b; Lefebvre and
Lumsden 1989, 1994). According to Lefebvre and Lumsdens (1994) formal
definition of the process, relexification builds a new lexicon in the following
way. The lexical entries of the lexicons of the substratum languages are copied,
and the phonological representations in these copied lexical entries are replaced
with phonological representations derived from the phonetic strings of the
superstratum language or by null forms. The second step is referred to as
relabelling. The choice of the pertinent phonetic string in the superstratum language to relabel a copied lexical entry is based on their use in specific semantic
and pragmatic contexts such that, as is advocated in Muysken (1981), the
semantics of the superstratum string must have something in common with the
semantics of the substratum lexical entry that is being relabelled.
In the literature on creole genesis, it has been pointed out that the makers
of a creole do not identify the functional category lexical entries (i.e. determiners, complementizers, tense, mood and aspect markers, etc.) of the superstratum language (cf. Lefebvre 1984; Carden and Stewart 1988; Mufwene 1991;
Lefebvre and Lumsden 1994; etc.) because of the limited access that they have
to the data (cf. e.g. Thomason and Kaufman 1991). In Lefebvre and Lumsden
(1994), it is claimed that the functional category lexical entries copied from the
substratum languages are relabelled on the basis of phonetic strings of superstratum lexical categories. It is further claimed that when relexifiers do not
find any appropriate phonetic string in the superstratum language, that is, a
form which is both semantically and distributionally suitable, the copied
functional category lexical entry may be assigned a phonologically null string,
such that when this lexical entry is used in an utterance, it is not pronounced.
The formal representation of the process of relexification provided in Lefebvre
and Lumsden (1994) is illustrated schematically in (1).4

. Relexification

Lumsden 1994). The data resulting from relexification will be shown to feed
the processes of dialect levelling and reanalysis. In this chapter, I illustrate how
these three processes apply in creole genesis on the basis of Haitian data
involving functional category lexical entries. Before doing so, I provide a
definition of these three processes.3

Relexification in Creole genesis

creole
[phonology]j or []
[semantic feature]k
[syntactic feature]n

[phonetic string]j used


in specific semantic
and pragmatic contexts

target language

The process of dialect levelling has been observed in situations where dialects
or languages are in contact (cf. e.g. Trudgill 1986; Siegel, this vol.). In Lumsden and Lefebvre (1994), it is proposed that this process plays a role in the
development of creole languages. Since relexification is a mental process, it is
necessarily an individual activity. Typically, situations where creoles are created involve several substratum languages (cf. Whinnom 1971). Thus, although
relexification from a single superstratum language provides the early creole
community with a common vocabulary (cf. Lumsden and Lefebvre 1994), the
relexified lexicons from different substratum languages would not be homogeneous in the early creole. As Lumsden and Lefebvre (1994) proposed, some of
these differences might be levelled out by the process of dialect levelling.

. Dialect levelling

The data presented in this chapter will be shown to support the above claims.

Since relexification is the first step in second language acquisition, the original
aim of the relexifiers is to reproduce the phonetic strings of the superstratum
language (. . .) Since the relexifiers intend to reproduce the phonetic strings of the
superstratum language, the creole word order for lexical categories will be the
word order of lexical categories in the superstratum language (. . .). On the other
hand, since the relexifiers do not identify the superstratum functional categories,
the word order for creole functional categories will be the same as the word order
of the substratum categories that they were relexified from.

Lefebvre and Lumsdens (1994: 10) proposal makes the following prediction
concerning word order in creoles:

(=(1) in Lefebvre and Lumsden 1994)

(1) original lexical


entry
[phonology]i
[semantic feature]k
[syntactic feature]n

Claire Lefebvre

This section illustrates how relexification applies to functional category lexical


entries during the period when native speakers of substratum languages are
targeting the superstratum language. Cases where the copied lexical entry is re-

. Phase : Adult native speakers of substratum languages target


the superstratum language

In light of this theoretical background, I will now show how these three
processes are implemented in the genesis of Haitian creole. I begin with phase
1 when adult native speakers of different substratum languages target the
superstratum language and use the mental process of relexification in order to
create a new language. Then I illustrate the processes that apply in the second
developmental phase of the creole when the speakers target the relexified
lexicons, that is, the early creole.

In the absence of visible phonological signals for a particular functional category,


speakers [of the early creole] use periphrastic constructions to clarify information
that is not being signalled. Speakers will then copy the phonological form of the
key element of the periphrastic construction into the lexical entry of the previously hidden functional category. In this way, reanalysis provides an explicit signal
for a creole lexical entry that was generated by relexification but did not acquire
a phonological signal through relabelling . . . When reanalysis assigns this lexical
entry a phonological signal, so that it becomes explicit in the creole expression,
the creole lexical entry is visibly more like that of the substratum language.

Reanalysis is a process through which a particular phonological form associated


with one lexical entry comes to be associated with another lexical entry
(Lightfoot 1979). This process, sometimes referred to as grammaticalization (cf.
e.g. Sankoff 1990; Hopper and Traugott 1993), has been shown to play a role in
cases of regular linguistic change. For example, the preposition of in English has
been reanalyzed as a case marker (Chomsky 1981). Likewise, according to
Kaynes (1981) analysis, the French forms and de have a double status as
prepositions and as complementizers. Lefebvre and Lumsden (1994) propose
that, when speakers target the speech of the creole community, that is, the early
creole, reanalysis plays a role in the further development of the creole. They
claim, however, that in the early creole this process applies to a lexical entry
that has been created through relexification but assigned a phonologically null
representation (cf. (1)). As Lefebvre and Lumsden (1994: 13) put it:

. Reanalysis

Relexification in Creole genesis

Haitian

Haitian

l enfant
det child
the child
b. le livre
det book
the book
c. la table
det table
the table

(3) a.

French

In contrast, the French determiner appears before the noun, as shown in (3),
and it is specified for gender and number. Le is masculine singular, la is feminine singular, les is plural, and l is a phonologically conditioned allomorph.

timounn nan
child
det
the child (in question/that we know of)
b. liv la
book det
the book (in question/that we know of)

(2) a.

.. The [+definite] determiner


This section argues that the lexical entry of the Haitian determiner has been
created through the process of relexification. The data and analysis reported
on in this section are drawn from Lefebvre (1994a), based on a series of papers
on the various facets of this Haitian lexical entry (e.g. Lefebvre 1982, 1992,
1996a, 1998a; Lefebvre and Massam 1988).
Haitian creole has a postnominal determiner la (with the phonologically
conditioned allomorphs a, an, nan and lan), as illustrated in (2). The presence
of this determiner indicates that the information conveyed by the noun phrase
is part of the shared knowledge of the participants in the conversation (cf.
Fournier 1977; Lefebvre 1982; Lefebvre and Massam 1988). The Haitian
determiner is not marked for gender.

. Copy and relabel on the basis of superstratum phonetic strings

labelled on the basis of a superstratum phonetic string and cases where the
copied lexical entry is assigned a phonologically null form will be discussed in
turn.

Claire Lefebvre

French

(5) *(Le) pain est bon pour la sant


(det) bread is good for det health
Bread is good for ones health. (= (24) in Lefebvre 1994a)

French

Haitian

Finally, in Haitian, the head noun and the determiner may be separated by a
relative clause, as in (8).

(7) Jan manje pen.


John eat
bread
John ate bread.

According to Haase (1965), this partitive determiner has been attested in


French since the fifteenth century. Milner (1978: 24) points out the exceptional character of French with respect to this partitive determiner and notes
that in most languages the determiner does not appear in contexts where the
French partitive determiner is manifested. Haitian follows the pattern of the
majority of languages, as shown in (7), the Haitian counterpart of the French
sentence in (6).

(6) Jean a mang du pain.


John aux eat
de+le bread
John ate bread. (=(25) in Lefebvre 1994a)

Furthermore, French has a partitive determiner de+la or du (a contracted form


of de + le), which appears with mass nouns, as in (6).

Haitian

(4) Pen bn pou lasante.


bread good for health
Bread is good for ones health. (= (19) in Lefebvre 1994a)

In contrast with the Haitian determiner, the French determiner does not necessarily identify old or known information. According to Milner (1978: 23),
the definite determiner is either anaphoric, identifying an object that already
has been mentioned, or cataphoric. In the latter case, larticle annonce une
relative ou un gnitif sans quaucune mention antrieure ne soit requise.
The Haitian determiner cannot appear with nouns that have a generic or
mass interpretation, but the French determiner must appear with such nouns
(cf. Milner 1978: 25). These facts are illustrated in (4) and (5), respectively.

d. les livres/tables
det books/tables
the books/tables

Relexification in Creole genesis

Haitian

French

river
home
history
world
bird
onions

<
<
<
<
<
<

la rivire
la case
lhistoire
la terre
les oiseaux
les oignons

the river
the house
history
the earth
the birds
the onions

Corresponding French DPs

Corresponding French DPs


de leau water
du feu
fire
du riz
rice

The data in (10) and (11) further show that the creators of Haitian did not
identify the French determiners as independent morphemes and often analyzed them as part of the phonetic strings of the nouns with which they
appear. Further support for this claim comes from the fact that Haitian nouns
which contain an agglutinated French determiner may occur with the
postnominal Haitian determiner, as shown in (12).

(11) Haitian nouns


dlo water <
dife fire
<
diri rice <

The French partitive determiner illustrated in (6) above is also found as part
of Haitian simple nouns, as shown in (11).

larivy
lakay
listwa
lat
zwazo
zonnyon

(10) Haitian nouns

The determiner in Haitian creole and the determiner in French thus have
quite different semantic and syntactic properties, which indicates that the
properties of the Haitian creole determiner are not derived from French.
Moreover, the French determiner does not appear to have been the source of
the phonological representation of the Haitian determiner either. The French
determiner is often found as part of Haitian simple nouns, as shown in (10).

(9) *Le [qui est parti] homme


det [who aux leave] man
Lit.: The [who left] man (=(26) in Lefebvre 1994a)

By contrast, in French, the head noun and the determiner may not be separated by a relative clause, as shown by the ungrammaticality of the sentence
in (9).

(8) Mounn [ ki
pati] a.
man
op re-pro leave det
The man who left. (=(20) in Lefebvre 1994a)

Claire Lefebvre

Haitian

Haitian

Fongbe

Fongbe

Fongbe

As is the case in Haitian, Fongbe has no partitive determiner and in Fongbe


(cf.(15)), as in Haitian (cf. (7)), bare nouns are interpreted as mass nouns.

(14) W` xx ny n n lnm` yn.


bread good for health
Bread is good for ones health. (= (19) in Lefebvre 1994a)

This determiner, like the Haitian determiner, indicates that the information
conveyed by the noun phrase is part of the shared knowledge of the participants in the conversation (cf. Lefebvre 1992). Thus, like the Haitian determiner (cf. (2)), the Fongbe determiner in (13) is obligatorily anaphoric.
The Fongbe determiner cannot appear in noun phrases that have a
generic or a mass interpretation, as shown in (14), which parallels the Haitian
data in (4).

child det
the child (in question/that we know of)
b. wm n
book det
the book (in question/that we know of)

(13) a.

Consequently, like Sylvain (1936) and Fournier (1977), I conclude that the
French determiner could not have been the phonetic source of the Haitian
determiner.
Where, then, do the properties of the Haitian determiner come from? I
argue below that, while the semantic and syntactic properties of the Haitian
determiner are derived from the substratum languages corresponding lexical
entry, its phonological representation is derived from the superstratum lexical
item l which occurs at the end of constituents.
For example, Fongbe, one of the substratum languages of Haitian (cf.
Lefebvre 1993a), has a postnominal determiner (with a phonologically determined allomorph n), as illustrated in (13).

larivy a
river det
the river (in question) (=(29) in Lefebvre 1994a)
b. diri a
rice det
the type of rice (in question)

(12) a.

Relexification in Creole genesis

Fongbe

Haitian
Fongbe

Fongbe

Furthermore, in both Haitian and Fongbe, the determiner is required in the


syntactic structure of relative (cf. Koopman 1982; Lefebvre 1986), conditional
and factive clauses (cf. Lefebvre and Massam 1988; Kinyalolo 1993; Collins
1994). Finally, in both Haitian and Fongbe, the determiner plays a central role
in the structure of the clause (cf. Lefebvre 1992, 1996a, 1998a), and for some
speakers, it may appear in constructions involving verb doubling phenomena
(cf. Lefebvre and Ritter 1993; Lefebvre 1994b).
The above data show that the semantic and syntactic properties of the
Haitian creole determiner are strikingly similar to those of the Fongbe determiner. Both languages contrast with French in the same way with respect to
this functional category lexical entry. It thus seems that substratum languages
of the type of Fongbe were the source of the properties of the Haitian determiner. Following the analysis in Lefebvre and Massam (1988) and Lumsden
(1989, 1991) for Haitian, and the analysis in Brousseau and Lumsden (1992)
for Fongbe, I assume that the forms la and (and their allomorphs) are the
head of the functional category phrase DP. These determiners are specified for
the feature [+definite] and they take their complement to the left.
What, then, is the source of the phonological form of the Haitian creole
determiner? In the literature, it has been proposed that the French form l
is a good candidate for the phonetic source of the Haitian determiner la (cf.
Faine 1937; Goodman 1964; Fournier 1977; Valdman 1978; etc.). In addition

(17) Sn [
y]
man op re-pro leave det
The man who left.

In Fongbe (cf. (17)), as in Haitian (cf. (8)), the head noun and the determiner
may be separated by a relative clause.

(16) diri
a
m` lk
rice
det
the type of rice (in question)

In both languages, a determiner that co-occurs with a mass noun yields the
interpretation type of x, as shown below.

(15) K` k bl
Koku eat bread
Koku ate bread.

Claire Lefebvre

Cet homme -l
vient darriver.
dem man
[+deic] come arrive
This/that man just arrived.

French

These examples show that the French form l has a distribution that is similar
to the distribution of the Fongbe determiner and, furthermore, the deictic
interpretation of this French form (cf. (18)) overlaps with the interpretation of
the Fongbe determiner. Thus, the French form l, pronounced [l] or [l] in
popular dialects of French, was probably the source of the phonological form
of the Haitian determiner la. As is pointed out in Lefebvre (1994a, 1996a) the
syntactic and semantic properties of this French form are not the same as
those of the Haitian determiner. The French l can appear with definite or
indefinite noun phrases (cf. (19a, b)), whereas the Haitian determiner is
incompatible with indefinite noun phrases (cf. (4)). Furthermore, when it is
used in the context of a clause (cf. (19c)), l has no semantics in common with
the Haitian determiner. Thus, the French form l may have been the source of
the phonological representation of the Haitian creole determiner but it did not
contribute the syntactic and semantic properties of this functional category
lexical entry.
The creation of the lexical entry for the Haitian determiner thus proceeded
as depicted in (20), which should be read in light of the general schema in (1).
The copied lexical entry corresponding to the Fongbe determiner was
relabelled on the basis of the phonetic matrix of the French form l, yielding
the Haitian determiner la.

L homme l vient darriver.


Popular French
det man
int come arrive
The man, uh, just arrived.
b. Un homme l vient darriver.
Popular French
det man
int come arrive
A man, uh, just arrived.
c. Un/l homme l, qui vient darriver l . . .
Popular French
det man
int who come arrive int
A/the man, uh, who just arrived, uh. . . (= (30) in Lefebvre 1994a)

(19) a.

(18)

to its use as a locative adverb meaning there, the French l may appear
following the noun in a noun phrase as an emphatic deictic marker, as in (18),
or as a simple interjection after a noun phrase or a clause, as in (19). Note that
the use of l shown in (19) has been attested to exist in French since at least
the seventeenth century (cf. Fltre 1970: 141, 146; Hull 1975: 2).

Relexification in Creole genesis

French
[l]/[l]
used in specific
semantic and pragmatic
contexts
Haitian Creole lexical entry
/la/
[+definite]
complement/head

Haitian

Haitian

Furthermore, sa/sila may occur within the same nominal phrase as the
possessive marker, the [+definite] determiner, and the plural marker. This is
shown in (22).

bag sa //
sila
ring [+deic] [prox]
this/that ring
b. mounn sa // sila
man [+deic] [prox]
this/that man (=(2) in Lefebvre 1997)

(21) a.

.. The [+deictic] terms


The [+deictic] terms of Haitian present yet another case of functional category
lexical entries relexified by the creators of Haitian. This section summarizes
the findings of an extensive analysis of these terms in Lefebvre (1997, 1999
and 2001).
In Haitian, there are two demonstrative terms sa and sila (cf., among
others, Goodman 1964: 50; Sylvain 1936). In the grammar of the Haitian
speakers who provided me with the data discussed below, while sila is used only
to point at objects that are far from the speaker, sa is used as a general deictic
term to point at objects that are either close to or far from the speaker. So,
while sila is specified as [proximate], sa is not specified for any value of the
feature [" proximate]. In the examples below, the meaning of sila is rendered
as that and the meaning of sa is rendered as this/that. (Other interpretive
patterns will be discussed at the end of this section.) Furthermore, as is pointed
out in Valdman (1978), these Haitian demonstrative terms may be used to
point at objects which are either animate and inanimate, as illustrated in (21)
below; these terms are thus not specified for the feature [" animate]. These
[+deictic] determiners appear postnominally, as shown in (21).

(20) Fongbe lexical entry


/ /
[+definite]
complement/head

Claire Lefebvre

Haitian

Haitian

(=(9) in Lefebvre 1997)

French

French

The semantic and syntactic properties of the lexical entries ce(t), cette and ces
are presented in (25).

this/that
ring(s)
1these/those2

ce
garon(s)
1ces2
this/that
boy(s)
1these/those2
b. cette
bague(s)
1ces 2

(24) a.

French has a series of demonstrative terms which can be divided into three
major groups on the basis of their categorial status: they are either determinative, pronominal, or adverbial. The forms ce(t), cette, ces constitute the first
group of French deictic terms. These forms are nominal determiners and they
are mutually exclusive with other determiners. They agree in gender and
number with the noun they determine, so they bear gender and number
features: ce(t) (m. sing.), cette (f. sing.), ces (pl.). As is shown in (24), these
forms occur prenominally. They can be used with animate (24a) or inanimate
(24b) objects. Therefore, they are not specified for any value of the feature ["
animate]. Furthermore, they can be used to point at an object that is either
close to or far from the speaker; thus, they are not specified for any value of
the feature [" proximate].

/sa/
this/that
[+deictic]
b. /sila/
that
[+deictic]
[proximate] (=(8) in Lefebvre 1997)

(23) a.

Since sa and sila are determiners, they are not identified for the categorial
features [+N, V]. The semantic and syntactic properties of the lexical entries
sa and sila are presented in (23).

(22) bag mwen sa //


sila a yo
Haitian
ring poss [+deic] [prox] det pl
these/those rings of mine (in question/that we know of) (=(3) in
Lefebvre 1997)

Relexification in Creole genesis

French

French

French

French

a: /sa/ this/that
[+deictic]
[+N, V]
[animate]
b. cela: /sla/ this/that
[+deictic]
[+N, V]
[animate]

(27) a.

French

French

The semantic and syntactic properties of the lexical entries a, cela and ceci are
summarized in (27).

(26) J ai vu a
// cela
// ceci.
I aux see [+deic] // [+deic] // [+prox]
I saw this/that // this/that // this. (=(14) in Lefebvre 1997)

The properties of the French lexical entries in (25) differ from those of the
Haitian lexical entries in (23). Thus, it is unlikely that the properties of the
Haitian terms could be derived from these French determiners.
The second class of French demonstrative terms has two sets of pronominal forms, which are distinguishable on the basis of animacy. The three
pronominal forms a, cela, and ceci make up the first set (cf. (26)). Since these
forms may only be used with inanimate objects, they must be specified for the
feature [animate]. a and cela are general deictic terms which may be used
to point at objects that are either close to or far from the speaker; they are not
specified for any value of the feature [" proximate]. In contrast, ceci is used
only to point at objects close to the speaker. Consequently, it must be specified
for the feature [+proximate].

ce(t): /s/ ~ /st/ this/that


[+deictic]
[fem]
[pl]
b. cette: /st/ this/that
[+deictic]
[+fem]
[pl]
c. ces: /st/ these/those
[+deictic]
[+pl] (=(12) in Lefebvre 1997)

(25) a.

Claire Lefebvre

ceci: /ssi/ this


[+deictic]
[+N, V]
[animate]
[+proximate] (=(17) in Lefebvre 1997)

French

celui: /sli/ this/that


[+deictic]
[+N, V]
[fem]
[pl]
b. ceux: /s/ these/those
[+deictic]
[+N, V]
[fem]
[+pl]
c. celle(s): /sl/ this/that; these/those
[+deictic]
[+N, V]
[+fem]
(=(20) in Lefebvre 1997)

(28) a.

French

French

French

The properties of the French forms in (27) are very different from those of the
Haitian forms in (23). The French forms are pronominal and are specified as
[animate], whereas the Haitian forms are not pronominal and are not
specified for any value of the feature [" animate]. Furthermore, while the
paradigm of French pronouns in (27) has a form which lexically encodes the
feature [+proximate], the paradigm of Haitian forms in (23) has a form which
lexically encodes the feature [proximate]. Thus, a, ceci and cela are unlikely
to be the source of the semantic and syntactic properties of the Haitian
demonstrative terms.
The forms celui, ceux and celle(s) constitute the second set of demonstrative pronouns in French. These pronouns are used to point at objects that are
either animate or inanimate, and therefore, they are not specified for any value
of the feature [" animate]. These forms are specified for gender and number:
celui (m. sing.), ceux (m. pl.), celle(s) (f. sing./pl.). These demonstrative pronouns are neutral with respect to the feature [" proximate]. The semantic and
syntactic properties of the lexical entries celui, ceux, and celle(s) are shown
in (28).

c.

Relexification in Creole genesis

French

French

(=(23) in Lefebvre 1997)

French

French

Like the Haitian demonstrative terms, these two French lexical items occur postnominally (cf. (29) and (30)). They differ, however, with respect to categorial
features. Furthermore, in French, the positive value of the feature [" proximate]

l: /l/ ~ /l/ there/here


[+deictic]
adv
b. ci: /si/ here
[+deictic]
[+proximate]
adv

(31) a.

As shown in the above examples, l is a general deictic form, not specified for
any value of the feature [" proximate]. In contrast, ci is used to point at
objects that are close to the speaker; it is specified for the feature [+proximate]. The properties of the lexical entries l and ci are represented in (31).

J ai vu celui- ci.
I aux see [+deic] [+prox]
I saw this one.
b. J ai vu celui- l.
I aux see [+deic] [+deic]
I saw this/that one. (=(21) in Lefebvre 1997)
(30) a. cette
bague-ci
[+deic] ring [+prox]
this ring
b. cette
bague-l
[+deic] ring [+deic]
this/that ring (=(22) in Lefebvre 1997)

(29) a.

The French forms in (28) share only one property with the Haitian forms in
(23): none are specified for a value of the feature [" animate]. Aside from this,
the other properties of the French forms in (28) contrast with those of the
Haitian forms in (23). Thus, it seems that the semantic and syntactic properties of the Haitian deictic terms sa and sila are not derived from those of the
French forms celui, ceux, and celle(s).
Finally, there is a third group of deictic terms in French: the adverbials l
and ci. These two forms combine with the deictic pronominal forms of the
second group, as in (29), or with the deictic determiners occurring in nominal
structures, as in (30).

Claire Lefebvre

Fongbe

Fongbe

Since l and n occur as part of the determiner system of nouns, they are not
specified for the categorial features [+N, V]. The semantic and syntactic
properties of the lexical entries l and n are presented in (34).

(33) l` k ce
l //
n
l
Fongbe
ring poss [+deic] [prox] det pl
these/those rings of mine (in question/that we know of) (=(25) in
Lefebvre 1997)

The [+deictic] determiners may occur within the same nominal phrase as the
possessive marker, the [+definite] determiner and the plural marker. This is
shown in (33), which parallels the Haitian data in (22).

l` k l
// n
ring [+deic] // [prox]
this/that // that ring
b. sn l
// n
man [+deic] // [prox]
this/that // that man (=(24) in Lefebvre 1997)

(32) a.

is lexically encoded, whereas in Haitian the negative value of the feature ["
proximate] is lexically encoded. Thus, it appears that these lexical items are not
the source of the properties of the Haitian deictic terms in (23).
How, then, did Haitian end up with the system described in (23)? The
properties of demonstrative terms in the substratum languages provide a clear
answer to this question: the properties of the creole lexical entries involving
demonstrative terms are derived from those of corresponding lexical items in
the substratum languages. For example, in Fongbe, as in Haitian, there are two
demonstrative terms: l and n (cf. Anonymous 1983). For the Fongbe
speakers whose grammar is discussed in this section, while n is used only to
point at objects that are far from the speaker, l is used as a general deictic
term for objects that are either close to or far from the speaker. So, while the
first form is specified as [proximate], the second form is not specified for any
value of the feature [" proximate]. In the examples below, the meaning of n
is rendered as that and the meaning of l is rendered as this/that. (Other
interpretive patterns will be discussed at the end of this section.) Furthermore,
the Fongbe demonstrative terms may be used with objects which are either
animate or inanimate. Thus, they are not specified for the feature [" animate].
As is the case in Haitian (cf. (21)), the Fongbe demonstrative terms occur
postnominally, as shown in (32).

Relexification in Creole genesis

(=(30) in Lefebvre 1997)

Fongbe

Fongbe

Haitian
/sa/
[+deictic]
b. /sila/
[+deictic]
[proximate]
a.

Fongbe
/l /
this/that
[+deictic]
/n /
that
[+deictic]
[proximate]
(=(31) in Lefebvre 1997)

As can be seen in (35), the semantic and syntactic properties of the Haitian
and Fongbe deictic terms are parallel; the Haitian lexical entries are different
from the Fongbe ones only in their phonological representations. This is
exactly what is expected under the relexification hypothesis.
How did the creators of Haitian establish a phonological form for the
copied lexical entries? In Lefebvre (1997) I argue extensively that the phonological form of the copied lexical entries was established on the basis of the
French form a, pronounced [sa], yielding sa in Haitian creole and cela, pronounced [sla] or [sl], yielding sila in Haitian creole. Alternatively, celui-l
might also have yielded sila in Haitian creole.
The creation of the lexical entries for Haitian demonstrative terms thus
proceeded as depicted in (36), which should be read in light of the general
schema in (1). The copied lexical entry corresponding to the general deictic
term l was relabelled on the basis of the phonetic matrix of the French

(35)

When we compare the Fongbe and Haitian data, we find a systematic parallel
between the properties of the demonstrative lexical entries in the two languages. In both languages, there is a paradigm of demonstrative terms with
two forms, both of which are determiners (cf. (22) and (33)). The distribution
of the two sets of forms is parallel. In both languages, the terms may be used
for either animate or inanimate objects, and in both languages, one term is
specified [proximate] while the other is unspecified for the feature ["
proximate]. Sila in Haitian and n in Fongbe are both [proximate], while sa
and l are [" proximate] (cf. (23) and (34)). In contrast with French, both
languages lack a form that lexically encodes the positive value of the feature
[" proximate]. The lexical entries of the demonstrative terms for the two
languages are compared in (35).

/l /
this/that
[+deictic]
b. /n /
that
[+deictic]
[proximate]

(34) a.

Claire Lefebvre

Fongbe lexical
entry
/l /
[+deictic]

[sla]/[sila] used in
specific semantic and
pragmatic contexts

French

Haitian creole lexical entry


/sila/
[+deictic]
[proximate]
(=(32) in Lefebvre 1997)

Haitian Creole lexical entry


/sa/
[+deictic]

[sa] used in specific


semantic and pragmatic contexts

French

I would like to end this section by making more precise the interpretive
patterns associated with the two Haitian and Fongbe deictic terms. The
interpretive pattern discussed in the examples above and summarized in (35)
is taken from Lefebvre (1997). The validity of the interpretation of Haitian sila
in (35) has been challenged by DeGraff (1999) who claims that, like sa, sila is
a general deictic term. In Lefebvre (1999), I show that the interpretive pattern
I report in Lefebvre (1997) is also found in the literature (e.g. Sylvain 1936;
tienne 1974). Furthermore, I show that the pattern described in (35) (from
Lefebvre 1997) is one of three interpretive patterns.
Prior to Lefebvre (1997), two interpretive patterns had been reported for
sa and sila: one where there is no distal distinction between the two terms
(e.g., Frre 1974: 103; Joseph 1988: 112; Valdman 1978: 194; Valdman et al.,
1981; see also DeGraff 1999), and one where there is a distal distinction
between them. In his comparative grammar of French based creoles, Goodman (1964: 51) writes: In Haiti [. . .], these two meanings [this and that]
are distinguished formally, sa this and sila that. (See also Hall 1953: 29).

b. Fongbe lexical
entry
/n /
[+deictic]
[proximate]

(36) a.

general deictic pronoun a [sa], yielding the Haitian general deictic term sa.
Similarly, the copied lexical entry corresponding to the Fongbe [proximate]
pronoun n was relabelled on the basis of the phonetic forms cela [sla] or
celui-l [sila], yielding the Haitian [proximate] demonstrative pronoun sila.

Relexification in Creole genesis

sa
[" proximate]
[+ proximate]
[" proximate]

sila
[" proximate]
[proximate]
[proximate]

(37) tv K` k t n
table Koku gen det
Kokus table

(= (9) in Lefebvre and Lumsden 1992)

Fongbe

The following facts involving the realization of genitive Case illustrate the
situation where a copied lexical entry has been assigned a phonologically null
form. The data and analysis are drawn from Lefebvre and Lumsden (1992). As
is extensively argued in Brousseau and Lumsden (1992), Fongbe has a
postnominal genitive Case marker tn, shown in (37).

. Copy and relabel by a phonologically null form

.. Conclusion
The derivation of the Haitian [+definite] determiner and [+deictic] determiners discussed in this section illustrates how relexification proceeds in the case
of functional categories when the speakers of the substratum languages are
targeting the superstratum language. The data further show that relexifiers use
lexical items of the superstratum language in order to relabel copied lexical
entries. Other cases of functional items have been argued to have a similar
derivation. A detailed discussion of these cases can be found in Lefebvre
(1993b, 1994a, 1996b, 1998b).

Note that the third pattern appears to be a compromise between the first two:
sa is [" proximate], as in the first pattern, and sila is [proximate], as in the
second one. In Lefebvre (in press) I show that the three interpretive patterns
found for Haitian sa and sila are also found for Fongbe l and n, respectively, thus providing support to the relexification account of the Haitian
deictic terms.

interpretive pattern a
interpretive pattern b
interpretive pattern c

Tinelli (1970: 28) points out two interpretive patterns: sa is sometimes


replaced by sila, with no syntactic change [. . .], and no clear modification of
meaning, except [my emphasis] for some speakers who distinguish the remote
deictic sila from the proximate sa [. . .]. Lefebvre (1997) reports a third
pattern where sa is a general deictic term and sila is [proximate]. These three
patterns are schematically represented below.

Claire Lefebvre

(= (10) in Lefebvre and Lumsden 1992)

French

(= (11) in Lefebvre and Lumsden 1992)

Haitian

The data discussed in this section show that the creators of Haitian did not
identify as such the functional categories of the superstratum language.
Instead, the creators of Haitian used the properties of their own functional
category lexical entries to create the creole lexical entries. This was achieved by
the process of relexification whereby the functional category lexical entries
were copied, and relabelled either on the basis of superstratum phonetic
strings or by a phonologically null form.5

. Conclusion

Lefebvre and Lumsden (1992) account for the apparent difference between
Haitian and both of its contributing languages as follows. They hypothesize
that the lexical entry associated with Fongbe tn has been copied but not
relabelled. Since Case is required by universal grammar (cf. Chomsky 1981;
Travis and Lamontagne 1992), Lefebvre and Lumsden (1992) assume that
there is indeed a functional category realizing Case in the Haitian noun phrase
in (39) and that this category is phonologically null. On this account, then,
speakers of a language like Fongbe who were relexifying their lexicon on the
basis of data from French would have relabelled the copied lexical entry of
their genitive Case marker with a phonologically null string. Crucially, the
Haitian phrase headed by the null form in (39) is argued to have the properties of a Genitive Phrase, as in (37), rather than those of an Objective one, as
in (38) (see Lumsden 1996).

(39) tab Jan an


table Jan gen det
Johns table

Finally, central and southern Haitian manifest no overt Case marker in the
context of a nominal argument, as shown in (39).

(38) la table de/ Jean


det table case Jean
The table of John

In French noun phrases, Case is realized as de or (cf. Kayne 1981). The


realization of Case precedes the noun, as shown in (38).

Relexification in Creole genesis

Haitian
Fongbe

(=(32) in Lefebvre 1994a)

Haitian
Fongbe

(42) krab la yo/*yo la


s n l /*l
crab det pl
the crabs (in question)

(=(33) in Lefebvre 1994a)

Haitian
Fongbe

In both Haitian and Fongbe, the plural marker may occur within the same
noun phrase as the determiner. In this case, at S-structure, the plural marker
follows the determiner.

(41) M achte krab.


N x
s n.
I buy crab
I bought (some) crabs.

A noun followed by the plural marker alone, is always interpreted as [+definite]. As shown in (40), it cannot be interpreted as [definite]. Indefinite
plural is not signalled, as shown in (41).

(40) krab yo
s n l
crab pl
The crabs
*(Some) crabs (=(31) in Lefebvre 1994a)

The process of dialect levelling is illustrated below on the basis of data involving the Haitian plural marker. The data are drawn from Lefebvre (1994a). As
shown in Lefebvre (1994a), the Haitian plural marker yo shares a number of
properties with the Fongbe plural marker l . For one thing, both plural
markers occur postnominally, as shown in (40).

. Dialect levelling

The second developmental phase of the creole starts when the creators of the
creole target the relexified lexicons, that is, the early creole. This situation gives
rise to two other processes: dialect levelling (cf. Lumsden and Lefebvre 1994)
and reanalysis (cf. Lefebvre and Lumsden 1994). These processes will be
discussed on the basis of data involving Haitian functional category lexical
entries.

. Phase : The creators of the creole target the relexified lexicons

Claire Lefebvre

Haitian
Fongbe

French
(=(36) in Lefebvre 1994a)

French

(=(37) in Lefebvre 1994a)

French

(46) a.

krab yo
crab pl
the crabs

Haitian

Thus, so far there is a systematic parallel between Haitian yo and Fongbe l ,


which both contrast with the expression of the category number in French.
There is a difference, however, between Haitian and Fongbe, in that the
Haitian plural marker is also used as a third person plural personal pronoun,
as in (46), while the Fongbe plural marker does not fulfill this function. In
Fongbe, the third person plural personal pronoun is rendered by a different
lexical item, as shown in (47).

(45) Jean a mang des


pommes.
John aux eat
de+les apple
John ate (some of the) apples.

Second, in contrast with Haitian, in French, the expression of plural is obligatory even when a noun is indefinite. This is shown in (45) where des is the
contracted form of de + les (cf. Milner 1978, for a discussion of the French
partitive de).

Jean a mang la pomme.


John aux eat
the apple
John ate the apple.
b. Jean a mang les pommes.
John ate the apples.

(44) a.

The properties of the Haitian plural marker discussed so far are similar to
those of the corresponding marker in Fongbe. These properties contrast
systematically with the expression of number in French. First of all, in French,
the feature [plural] is an obligatory feature of the prenominal determiners
and is expressed as part of the determiner. While le and la are the singular
forms of the definite determiner, les is the plural form of the determiner.

(43) krab la
s n
crab det num
the crab (in question) (=(34) in Lefebvre 1994a)

As shown in (43), in these languages, the singular is not indicated by a specific


marker.

Relexification in Creole genesis

Fongbe

Fongbe

Haitian

(=(40) in Lefebvre 1994a)

French

The French pronoun eux thus has a similar surface distribution to the Haitian
form yo. It is clear, however, that it shares only a few semantic features with yo
since, in French, eux is only pronominal; unlike Haitian yo, it is never used as
a plural determiner. These facts suggest that, if the Haitian form yo derives its
phonological representation from the phonetic matrix of the French form eux,
it does not derive its semantic properties from that form. How did Haitian yo

Les gars, eux, ils . . .


French
det guys them, they . . .
The guys, them, they . . .
b. Les filles, eux, ils . . .
Popular french
det girl them they . . .
The girls, them, they. . . (=(41) in Lefebvre 1994a)

(49) a.

Furthermore, in colloquial French, the third person pronoun eux may also
occur as an emphatic pronoun at the end of a nominal constituent, as shown
in (49) (A. M. Brousseau, p.c.). Note the neutralization in gender shown in
(49b), also noted by Gougenheim (1973).

(48) Eux, ils mangent du riz chaque jour.


them, they eat
part rice every day
Them, they eat rice every day.

How can the difference between the two languages be accounted for within the
framework of the relexification hypothesis?
Sylvain (1936) and Goodman (1964) have suggested that the use of yo as
a plural marker is in fact an extended usage of the third person plural pronoun. In their view, the phonological representation of yo is derived from the
strong form of the third person personal pronoun eux. The pronoun eux in
French occurs as an emphatic form before a clitic, as in (48).

s n l
crab pl
the crabs
b. Ye y.
3pl leave
They left.

(47) a.

b. Yo pati.
3pl leave
They left.

Claire Lefebvre

D
[+denite]

NumP

Num
ti

DP

Third person pronoun

yoi

On this account then, both the plural marker and the third person plural
pronoun are of the category Num. Whereas in Haitian creole, Yoruba,
Mandingo, etc., there is only one lexical entry fulfilling these two functions, in
Fongbe the category Num is realized by two different morphemes: l in the
context of a noun phrase and ye elsewhere. It thus appears that, in this case,
Haitian follows the pattern of substratum languages other than Fongbe.
Lumsden and Lefebvre (1994) propose the following account of the
historical derivation of Haitian yo within the framework of the relexification
hypothesis. They assume that, in languages of the type of Ewe, there is only
one lexical entry wo whereas, in languages like Fongbe, there are two pertinent
lexical entries (i.e. ye and l ). In their view, the relexification hypothesis
predicts that there would be at least two basic dialects in the early creole with
respect to these lexical entries. Lumsden and Lefebvre (1994) hypothesize that
speakers of a language like Fongbe looked for two morphemes in the superstratum data, one to relexify the plural marker and the other to relexify the
third person plural pronoun. They further hypothesize that speakers of a

DP

(50) First/second person pronoun

come to be used both as a third person plural pronoun and as a plural marker,
while the corresponding form in French is only used as a pronoun and the
corresponding form in Fongbe is only used as a plural marker?
Interestingly enough, several West African languages present cases where
the third person plural pronoun also serve as a plural marker. For example, in
Ewe, wo is the form for both the third person plural pronoun and the plural
marker (cf. Westerman 1930: 45, 57). According to Goodman (1964: 46, 47),
Yoruba, Mandingo, and other languages also show this double function of a
single morpheme. The proposal in Ritter (1992) provides a theoretically
motivated account of this fact. She argues that, while first and second person
pronouns are of the category D(eterminer), third person pronouns are of the
category Num(ber). According to her analysis, third person pronouns are
derived by raising Num to D, as is schematically represented in (50).

Relexification in Creole genesis

Haitian

(52) M ap
ap sti.
I def-fut imp go-out
I will be going out.

(=(20) in Lefebvre 1996b)

Haitian

In Lefebvre (1996b), it is argued that there are two lexical entries with the
phonological representation ap in the Haitian lexicon: one encoding imperfective and one encoding future. The first argument supporting this claim comes
from data drawn from a subset of speakers who allow two aps within a single
clause, as shown in (52).

(51) Mari ap sti.


Mary go-out
Mary is going out./Mary will go out.

In this section, I illustrate how the process of reanalysis operates in the context
of the development of a creole on the basis of data pertaining to the tense,
mood and aspect (TMA) system of Haitian. The data are drawn from an
extensive analysis of the TMA system of Haitian, as compared to those of its
source languages (cf. Lefevre 1996b). The pertinent data involve the Haitian
preverbal marker ap. It is a well-known fact that a Haitian sentence containing
the preverbal marker ap is ambiguous between an imperfective and a future
reading (cf. Damoiseau 1988; Dchaine 1991; Lumsden 1996). This is shown
in (51).

. Reanalyis

language like Ewe looked for only one form, since their own lexicon provided
a single lexical entry fulfilling both of these purposes. Finally, Lumsden and
Lefebvre propose that the differences between the two types of grammar in the
early creole were reconciled through the process of dialect levelling. In this
particular case, the early creole dialect that came to dominate was the one that
used a single morpheme in both contexts. This analysis accounts for the
properties manifested by yo in modern Haitian creole.
The above discussion shows how the process of dialect levelling operates
on the basis of the early creole lexicons relexified from various substratum
languages. The role of this process in creole genesis accounts for the fact
that the properties of some specific lexical entries of the creole may depart
from those of corresponding lexical entries in the individual substratum
languages.

Claire Lefebvre

(54) Resultative verbs


Mari ap
w Jan.
Mari n
m Jan.
Mary def-fut see John
Mary will see John. (=(120) in Lefebvre 1996b)

(53) Dynamic verbs


Mari ap
prepare pat.
Mari n
a
w .
Mary def-fut prepare dough
Mary will prepare dough. (=(119) in Lefebvre 1996b)
Haitian
Fongbe

Haitian
Fongbe

A second argument is that speakers who do not accept two co-occurring aps
still have the pertinent interpretations. For example, for this second group of
speakers, a sentence containing ap such as M ap sti may be assigned three
interpretations: (a) I am going out, where ap is assigned an imperfective
reading; (b) I will go out where ap is assigned a future interpretation; or (c) I
will be going out. This latter interpretation shows that the second group of
speakers (those who do not pronounce two aps in a row6) still have the
interpretation corresponding to (52), which does contain two aps. Thus, the
two groups of speakers present similar interpretive data regardless of whether
they allow the co-occurrence of two aps at surface structure. These sets of facts
argue that there are two lexical entries for ap in the Haitian lexicon.
In Lefebvre (1996b), it is argued that the lexical entry of the imperfective
preverbal marker ap has been created by relexification. On this analysis, the
phonological form of this preverbal marker was established by relabelling on
the basis of the French preposition aprs, which occurs in the French periphrastic progressive, as in Marie est aprs manger Mary is eating. The lexical
entry of the future marker ap is also argued to have been created through
relexification but, in this case, it acquired its phonological representation
through the process of reanalysis. In the remainder of this section, I summarize the arguments supporting this claim. I begin by showing that the properties of the definite future marker ap in Haitian systematically parallel those of
the definite future marker n in Fongbe.
First, the definite future markers ap and n both situate, with respect to
the moment of speech, an event that is expected to definitely take place in the
near future. The event may coincide with the point of reference, yielding a
definite future interpretation with verbs of all aspectual classes, as shown
in (53)(55).

Relexification in Creole genesis

(=(121) in Lefebvre 1996b)

Haitian
Fongbe

(=(122) in Lefebvre 1996b)

Haitian
Fongbe

(=(123) in Lefebvre 1996b)

Haitian
Fongbe

These facts show that the definite future markers in Haitian and Fongbe have
similar morphological and semantic properties (cf. Lefebvre 1996b, for a more
detailed discussion).
According to the relexification hypothesis, the properties of ap parallel
those of n because they have been copied from the original lexical entry. How
did the form ap become the phonological representation of the copied lexical
entry? Suppose that ap comes from the French adverb aprs after occurring in
sentence-initial position (e.g. Aprs avoir mang, nous partirons. After eating,
we will leave). But ap could not have become the phonological representation
of the copied lexical entry through the process of relabelling. Although Fongbe
n shares some elements of meaning with the French preposition of posteriority aprs, the two lexical items do not occur in the same surface position.
Whereas n occurs between the subject and the verb, aprs occurs in sentenceinitial position. In Lefebvre (1996b), it is argued that ap has become the
phonological representation of the copied lexical entry through reanalysis of

(57) Mari te ap
prepare pat.
Mari ko n
a
w .
Mary ant def-fut prepare dough
Mary would prepare dough.
Mary would have prepared dough.

Interestingly enough, in both languages, a clause containing both the anteriority marker and the definite future marker is interpreted as conditional; this
holds for all three aspectual classes of verbs. An example containing a dynamic
verb is given in (57).

(56) Mari ap
deja
prepare pat.
Mari n
ko
a
w .
Mary def-fut already prepare dough
Mary will have prepared dough.

Or the event may precede the point of reference, yielding a future perfect
interpretation, as in (56).

(55) Stative verbs


Mari ap
knnn Jan.
Mari n
tu n
Jan.
Mary def-fut know John
Mary will know John.

Claire Lefebvre

Haitian

On the basis of Haitian functional category lexical entries, I have illustrated


how relexification, dialect levelling and reanalysis apply in the genesis and the
development of creole languages. It has been shown that the data produced by
the mental process of relexification feed the two other processes involved in the
further development of the creole: dialect levelling and reanalysis. The hypothesis that relexification plays a major role in creole genesis (cf. Lefebvre and
Lumsden 1994) provides a direct account of the division of properties observed
in the creole lexical entries. Creole lexical entries are predicted to have the same
semantic and syntactic properties as the correponding lexical entries in the
substratum languages, but with phonological representations taken from
phonetic strings of the superstratum language. The relexification hypothesis

. Conclusion

The following scenario is hypothesized within the framework, outlined in the


introduction, defining how reanalysis applies in creole genesis. The creators of
Haitian did not find any phonetic string in the superstratum language to
relabel the copied lexical entry corresponding to n. In phase 1, that is during
the period where they were relexifying their own lexicons, the relexifiers thus
assigned a null phonological form to the copied lexical entry (cf. (1)). They
used the sentence-initial preposition of posteriority apre after to clarify the
information which was not phonologically signalled. In phase 2, that is when
the agents of Haitian started targeting the relexified lexicons, the form apre
eventually made its way to the position between the subject and the verb. The
creators of the creole then copied the phonological form of this adverb of
posteriority onto the lexical entry of the previously phonologically null
functional category, the copied lexical entry corresponding to n in Fongbe. In
this way, reanalysis provided an explicit signal for a creole lexical entry that
was generated by relexification but was not assigned a phonological signal in
relabelling. When this lexical entry was assigned a phonological form, it had
the properties of the corresponding Fongbe lexical entry.

(58) Apre yo fin w-l


yo raknte . . .
After they finish see-him they told
After they had seen him, they told . . . (from Hall 1953: 221)

the Haitian adverb of posteriority apre (< French aprs after), which occurs
in sentence-initial position as shown in (58).

Relexification in Creole genesis

Adam, Lucien. 1883. Les Idiomes ngro-aryen et malo-aryen. Paris.


Alleyne, Mervyn C. 1981. Comparative Afro-American: An historical-comparative study of
English-based Afro-American dialects of the New World. Ann Arbor, MI: Karoma.
Anonymous. 1983. lments de recherche sur la langue Fon. Cotonou.

References

. For an account of this fact, see Lefebvre (1996b).

. Lefebvre and Lumsden (1992; 1994) also propose that a functional category lexical entry
that has been assigned a phonologically null form may be signalled by an overt form in the
specifier position of the projection headed by the null form. Due to space limitations, this
possibility will not be discussed here.

. The formal representation of the process of relexification in (1) is slightly different from
that in Muysken (1981). The motivation for these differences will be discussed elsewhere.

. Of course, I assume that, as is the case in other languages, innovation also plays a role in
creole genesis. This topic is discussed in Lefebvre (1998b).

. Quote from Sylvain: We are in the presence of a French that has been cast in the mould
of African syntax or . . . of an Ewe language with a French vocabulary.

. Quote from Adam: I go so far as to claim . . . that the so-called patois of Guyana and
Trinidad constitute Negro-Aryan dialects. By that I mean that the Guinean Negroes who
were transported to the colonies adopted the words of French but, as much as possible,
kept the phonetics and grammar of their mother tongues . . . Such a formation is clearly
hybrid . . . The grammar is no different from the general grammar of the languages of
Guinea.

* This chapter was written as part of the UQAM Haitian project financed by the SSHRC
and FCAR. I would like to thank Anne-Marie Brousseau, Kinyalolo Kasangati, Paul Law,
John Lumsden, Elizabeth Ritter, Pierrette Thibault and Raffaella Zanuttini for their very
useful comments on the work which led up to this chapter. I would also like to thank the
various speakers of Haitian and Fongbe who provided me with the data presented here.
Finally, many thanks to Andre Blanger for her help in formating the manuscript.

Notes

thus explains why creoles reflect the properties of both their superstratum and
their substratum source languages in the way they do (cf. Adam, 1883; Sylvain
1936; Goodman 1964; Koopman 1986; Huttar 1971; Keesing 1988, etc.) even
after several years of independent evolution. (For further discussion of the
issues discussed in this chapter see Lefebvre 1998b.)

Claire Lefebvre

Brousseau, Anne-Marie and John S. Lumsden. 1992. Nominal structure in Fongbe. In


Topics on the Syntax and Semantics of Fongbe, C. Lefebvre (ed.), 526. (Special issue of
the Journal of West African Languages, 22.1)
Carden, Guy and William A. Stewart. 1988. Binding theory, bioprogram, and creolization:
Evidence from Haitian Creole. Journal of Pidgin and Creole Languages 3.1: 168.
Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris.
Collins, Chris. 1994. The Factive Construction in Kwa. Travaux de recherche sur le crole
hatien 23: 3165. UQAM.
Damoiseau, Robert. 1988. lments pour une classification des verbaux en crole hatien.
tudes croles 11.1: 4164.
Dchaine, Rose-Marie. 1991. Bare sentences. In Proceedings of SALT I, S. Moore and
A. Wyner (eds.), 3150. (Cornell University Working Papers in Linguistics, 10.)
DeGraff, Michel. 1999. Empirical quicksand: Probing two recent articles on Haitian
Creole. Journal of Pidgin and Creole Languages 14: 359377.
tienne, G. 1974. Le crole du nord dHati: tude des niveaux de structure. Ph.D. dissertation, Universit des sciences humaines de Strasbourg.
Faine, Jules. 1937. Philologie crole: tudes historiques et tymologiques sur la langue crole
dHati. Port-au-Prince: Imprimerie de ltat.
Frre, Grard A. 1974. Haitian creole sound-system, form-classes, texts. Ph.D. dissertation,
University of Pennsylvania.
Fltre, Louis-Ferdinand. 1970. Le moyen picard, daprs les textes littraires du temps
(15601660): Textes, lexique, grammaire. Paris: Presses du Palais-Royal.
Fournier, Robert. 1977. N ap f yun ti-kose su la (La grammaire de la particule la en crole
hatien). Masters thesis, Universit du Qubec Montral.
Goodman, Morris F. 1964. A Comparative Study of Creole French Dialects. The Hague:
Mouton.
Gougenheim, Georges. 1973. Grammaire de la langue franaise au seizime sicle. Paris: A. et
J. Picard.
Haase, A. 1965. Syntaxe franaise du XVIIe sicle. Paris: Delagrave, 5e dition.
Hall, R. A. 1953. Haitian Creole: Grammar, Texts and Vocabulary. The American Anthropological Association. Memoir 74, Menasha Georges Banta Publishing Company.
Holm, John. 1988. Pidgins and Creoles. (Cambridge Language Surveys, 1). Cambridge:
Cambridge University Press.
Hopper, P. J. and E. C. Traugott. 1993. Grammaticalization. (Cambridge Textbooks in
Linguistics). Cambridge: Cambridge University Press.
Hull, A. 1975. On the Origin and Chronology of the French-Based Creoles. Paper presented
at the International Conference on Pidgins and Creoles, University of Hawaii.
Huttar, Georges L. 1971. Sources of creole semantic structures. Language 51.3: 68495.
Joseph, F. 1988. La dtermination nominale en crole hatien. Ph.D. dissertation, Universit
de Paris VII.
Kayne, Richard S. 1981. On certain differences between French and English. Linguistic
Inquiry 12.3: 349371.
Keesing, Roger M. 1988. Melanesian Pidgin and the Oceanic Substrate. Stanford, CA: Stanford University Press.

Relexification in Creole genesis

Kinyalolo, Kasangati K. W. 1993. Conditional and Factives in Fn. In Universit du


Qubec Montral. Research report 19921993, vol. iv: tudes syntaxiques, C. Lefebvre
and J. Lumsden (eds.), 161173.
Koopman, Hilda. 1982. Les constructions relatives. In Syntaxe de lhatien, C. Lefebvre,
H. Magloire-Holly and N. Piou (eds.), 167203. Ann Arbor, MI: Karoma.
Koopman, Hilda. 1986. The genesis of Haitian: Implications of a comparison of some
features of the syntax of Haitian, French and West African Languages. In Substrata
Versus Universals in Creole Genesis, P. Muysken and N. Smith (eds.), 231258. Amsterdam and Philadelphia: John Benjamins.
Lefebvre, Claire. 1982. Lexpansion dune catgorie grammaticale: le dterminant la. In
Syntaxe de lhatien, C. Lefebvre, H. Magloire-Holly and N. Piou (eds.), 2163. Ann
Arbor, MI: Karoma.
Lefebvre, Claire. 1984. Grammaire en contact: dfinition et perspectives de recherche.
Revue qubcoise de linguistique 14.1: 1149.
Lefebvre, Claire. 1986. Relexification in Creole genesis revisited: The case of Haitian
Creole. In Substrata Versus Universals in Creole Genesis, P. Muysken and N. Smith
(eds.), 279301. Amsterdam and Philadelphia: John Benjamins.
Lefebvre, Claire. 1992. AGR in languages without person and number agreement: The case
of the clausal determiner in Haitian and Fon. In Functional Categories, C. Lefebvre,
J. S. Lumsden and L. Travis (eds.), 137156. (Special issue of The Canadian Journal of
Linguistics 37.2).
Lefebvre, Claire. 1993a. The role of relexification and syntactic reanalysis in Haitian
Creole: Methodological aspects of a research program. In Africanisms in AfroAmerican Language Varieties, S. S. Mufwene (ed.), 254279. Athens, GA: The University of Georgia Press.
Lefebvre, Claire. 1993b. The role of relexification in Creole genesis: The case of functional
categories. Second International Symposium on Bilingualism: Grammatical aspects of
bilingualism. Hamburg.
Lefebvre, Claire. 1994a. Functional categories and relexification. Universit du Qubec
Montral. Paper read at the MIT Symposium on the Role of Relexification in Creole
Genesis: The Case of Haitian Creole. Research report ed. by C. Lefebvre and J. Lumsden. Montreal.
Lefebvre, Claire. 1994b. On spelling out E. Travaux de recherche sur le crole hatien 23:
130. Universit du Qubec Montral.
Lefebvre, Claire. 1996a. The functional category agreement and Creole Genesis. In
Creole Languages and Language Acquisition, H. Wekker (ed.), 153195. The Hague:
Mouton.
Lefebvre, Claire. 1996b. The tense, mood and aspect system of Haitian Creole and the
problem of transmission of grammar in Creole genesis. Journal of Pidgin and Creole
Languages 11.2: 231313.
Lefebvre, Claire. 1997. Relexification in Creole genesis: The case of demonstrative terms in
Haitian Creole. Journal of Pidgin and Creole Languages 12.2: 181201.
Lefebvre, Claire. 1998a. Multifunctionality and variation among grammars: The case of
the determiner in Haitian and in Fongbe. Journal of Pidgin and Creole Languages 13.1:
93150.

Claire Lefebvre

Lefebvre, Claire. 1998b. Creole Genesis and the Acquisition of Grammar: The Case of Haitian
Creole (Cambridge Studies in Linguistics, 88). Cambridge: Cambridge University
Press.
Lefebvre, Claire. 1999. On the empirical reliability of some Haitian data. Journal of Pidgin
and Creole Languages 14.2: 16.
Lefebvre, Claire. 2001. The interplay of relexification and levelling in Creole genesis and
development. Linguistics 39.2: 371408.
Lefebvre, Claire and John S. Lumsden. 1989. Les langues croles et la thorie linguistique.
In La crolisation, C. Lefebvre and J. S. Lumsden (eds.), 249272. (Special issue of The
Canadian Journal of Linguistics 34.3.)
Lefebvre, Claire and John S. Lumsden. 1992. On word order in relexification. Travaux de
recherche sur le crole hatien 10: 122. Universit du Qubec Montral.
Lefebvre, Claire and John S. Lumsden. 1994. Relexification in Creole genesis. Universit
du Qubec Montral. Paper read at the MIT Symposium on the Role of Relexification
in Creole Genesis: The Case of Haitian Creole. Research report, C. Lefebvre and
J. Lumsden (eds.). Montreal.
Lefebvre, Claire and Diane Massam. 1988. Haitian Creole syntax: A case for DET as head.
Journal of Pidgin and Creole Languages 3.2: 213243.
Lefebvre, Claire and Elizabeth Ritter. 1993. Two types of predicate doubling adverbs in
Haitian Creole. In Focus and Grammatical Relations in Creole Languages, F. Byrne and
D. Winford (eds.), 6591. Amsterdam and Philadelphia: John Benjamins.
Lightfoot, D. 1979. Principles of Diachronic Syntax. Cambridge: Cambridge University Press.
Lumsden, John S. 1989. On the distribution of determiners in Haitian Creole. Revue
qubcoise de linguistique 18.2: 6493.
Lumsden, John S. 1991. La distribution des modificateurs dans le syntagme nominal en
hatien. In La crolisation: Thorie et applications, A. Kihm (ed.), 4763. (Special issue
of Recherches Linguistiques de Vincennes, 20.)
Lumsden, John S. 1996. On the acquisition of nominal structures in the genesis of Haitian
Creole. In Creole Languages and Language Acquisition, H. Wekker, (ed.), 184205.
Berlin and New York: Mouton de Gruyter.
Lumsden, John S. and Claire Lefebvre. 1994. The genesis of Haitian Creole. Paper read
at the MIT Symposium on the Role of Relexification in Creole Genesis: The Case of
Haitian Creole. Research report, C. Lefebvre and J. Lumsden (eds.). Montreal.
Milner, J.-C. 1978. De la syntaxe linterprtation. Paris: ditions du Seuil.
Mufwene, Salikoko S. 1991. Pidgins, Creoles, typology and markedness. In Development
and Structure of Creole Languages, F. Byrne and T. Huebner, (eds.), 123144. Amsterdam and Philadelphia: John Benjamins.
Muysken, Pieter Cornelis. 1981. Half-way between Quechua and Spanish: The case for
relexification. In Historicity and Variation in Creole Studies, A. R. Highfield and
A. Valdman, (eds.), 5279. Ann Arbor, MI: Karoma.
Ritter, Elizabeth. 1992. Cross-linguistic evidence for number phrase. In Functional
Categories, C. Lefebvre, J. S. Lumsden and L. Travis (eds), 197219. (Special issue of
The Canadian Journal of Linguistics 37.2.)
Sankoff, G. 1990. The grammaticalization of tense and aspect in Tok Pisin and Sranan.
Language Variation and Change 2: 295312.

Relexification in Creole genesis

Sylvain, Suzanne. 1936. Le crole hatien: Morphologie et syntaxe. Wetteren: Imprimerie de


Meester; Port-au-Prince: Chez lauteur.
Thomason, Sarah Grey and Terrence Kaufman. 1991. Language Contact, Creolization and
Genetic Linguistics. Los Angeles: University of California Press.
Tinelli, Henri. 1970. Generative phonology of Haitian creole. Ph.D. dissertation, University
of Michigan.
Travis, Lisa and Greg Lamontagne. 1992. The case filter and licensing of empty K. In
Functional Categories, C. Lefebvre, J. S. Lumsden and L. Travis (eds.), 157174. (Special
issue of The Canadian Journal of Linguistics 37.2.)
Trudgill, P. 1986. Dialects in Contact. Oxford: Basil Blackwell.
Valdman, Albert. 1978. Le crole: Statut et origine. Paris: Klincksieck.
Valdman, Albert et al. 1981. Haitian CreoleEnglishFrench Dictionary. Bloomington, IN:
Indiana University Creole Institute.
Voorhoeve, Jan. 1973. Historical and linguistic evidence in favor of the relexification
theory in the formation of creoles. Language and Society 2: 133145.
Westerman, D. 1930. A Study of the Ewe Language. London: Oxford University Press.
Whinnom, K. 1971. Linguistic hybridization and the special case of Pidgins and Creoles.
In Pidginization and Creolization of Languages, D. Hymes (ed.), 91115. Cambridge:
Cambridge University Press.

Claire Lefebvre

You might also like