IMAGE 01h2LexDiffFin134.gif



disqualifies it from the status of a serial verb constructions in many (perhaps
most) analyses. It is not obvious that the sentence just cited is monoclausal.

As has been shown here under 9.1, the sentence in (16) does not contain "two subject
occurrences". And we wouldn't want to "disqualify" Tok Pisin sentences containing
predicate markers derived from erstwhile "subject occurrences", /i-/ < English <he>,
as in (40):


em i-wokabout i-go(Tok Pisin)
he PM-move PM-go
"he went away"
em i-kisim ol dispela pis i-kam long mi
he PM-take PL DEM fish PM-come to me
"he caught the fish and brought it to me"


from their "status of a serial verb construction".

IMAGE 01h2LexDiffFin05.gif
IMAGE 01h2LexDiffFin16.gif

Prepositional uses of verbs in the non-French atlantic creoles seem to me due to
accidental serialization, not to an earlier state where verb serialization was
productive in the sense of a process available to all verbs in the lexicon.

I have never claimed, and see no reason to assume that serialisation ever would have
been "available to all verbs in the lexicon".

However, that is what happens in African languages where serialization is an entirely
productive process:The semantic "decompression" or "unzipping" of complex
predicates into a transparent sequence of simple predicates (preserving the relevant
event-argument structure of the source predicate) must be available to all the verbs
in the lexicon. This is definitely not so for most "European-lexifier" creoles,
icluding the French ones. Parkvall's book comes nowhere to grips with typological
differences of this nature.

The book also focusses unduly on serial constructions which have no chance to ever
grammaticalizing in any SVO type of language. Take the competing structures of (41a)
and (41b) [open vowels underlined; DET /lá/ has a distribution determined by
pagmatics and cannot occur main-clause-finally]:

(41)aKòfí gbã fésrè lá kplé kpé
Koffi break window DET be-with stone
"Koffi broke the window with a stone"

(41)bKòfí tsókpé gbã fésrè
Koffi take stone break window
"Koffi broke the window with a stone"

(41a) has an erstwhile verb in the right syntactic place for serialization-induced
grammaticalization into a preposition. On the other hand, (41b) paraprases (41a)
with the fully verbal /tsó/ in a place where grammaticalization can hardly be
expected. Chance semantic decompression in any language can produce accidentally
structures of the type of (41b) as demonstrated by the Magoua example in (16).



IMAGE 01h2LexDiffFin41.gif
IMAGE 01h2LexDiffFin06.gif

All varieties have "bimorphemic" interrogatives (ki sa k i màlàd-màlàd? "who's
being sick like that?") [...]

Bimorphemic interrogatives are certainly (marginally) existant in prety much any
European language, but I have never seen any variety of colonial French (unlike
pidgins and creoles not co-existing with their lexifiers) that make _exclusive_ use
of such interrogatives.

Neither the relevant African languages nor the various varieties of creole French are
consistently bimorphemic as to their interrogatives (cf. Muysken & Smith 1990,
Wittmann 1987b). As for Haitian Creole, there are least six monomorphemic
interrogatives I can think of:kilès "which (one)", kote "where", kouman "how",
konben "how many", sa "what", witi "where". The situation gets worse when it comes

IMAGE 01h2LexDiffFin01.gif


Printed for Henri Wittmann <>

IMAGE 01h2LexDiffFin01.gif



to indirect interrogatives such as "when" with a multitude of monomorphemic
equivalents:lè, lò, kan. kou, kon, kòm. Magoua, on the other hand, has only one
monomorphemic interrogative:konben "how many". All the other interrogatives,
whether direct or indirect, subject or object, are not monomorphemic:ki-sà-k "who
(animate)", kò-sà-k "what (inanimate), kan-sà-k "when", kman-sà-k "how",
you-sà-k/ké-bor-k "where", kò-fér "why". Except for ké-bor-k and kò-fér, forms are
based on a transparent -sà-k (which can contract to -s-k or -sà). As for the i/é
alternation in ki-sà-k, ké-bor-k, I've already indicated elsewhere that i/é
neutralizes to open high [I] in surface phonology which no doubt gave rise to Creole
ki (cf. Wittmann 1987b, 1996).



IMAGE 01h2LexDiffFin178.gif
IMAGE 01h2LexDiffFin41.gif

The koine-creole continuum of the 17th-18th century underlying the structural
continuities of koine and creole varieties of colonial French has surviving
geographical continua relics in Louisiana and Saint-Barthelemy isle. The latter
situation in particular with its transitions from a White creole in the east to a
White koine in the west, by the very absence of any African input, constitutes an
unsurmountable enigma to all those to whom the concept of "creolization" is a
theoretical comfort they can't do without.

Not at all. For one thing, no one suggests that creolisation needs to involve
Africans. Secondly, the St. Barth case has convincingly been shown by Julianne Maher
to be the result of language import, the creole having been imported from the
Windward islands (via St. Vincent).

All forms of French in the Americas are ultimately "imported". Maher's original
hypothesis was that the colonists that founded Saint-Barth in 1648 already knew their
creole and that this could hardly be surprising at a time where creole must have been
already the vehicular language of the Karipuna. Maher's subsequent "Bernier
hypothesis" suggests importation of creole from Saint Vincent. The major problem
with this hypothesis is that no French creole has ever been in use on Saint Vincent.
When the isle was turned over to the English in 1763 (which must have been the time
when François Bernier and his two sons returned to Saint Barth), the entire Black
population was Karipuna speaking. This didn't change until 1797 when the entire
Black Karipuna population was deported by the British. At that time, the Black
Karipuna constituted 86% of the total population of 5800, the remaining 15% having
arrived mostly after 1763, except for the remaining Red Karipuna. Since Parkvall
agrees that creolization doesn't need to involve Africans, we may savely conclude
that koine and creole varieties of French were in use on Saint Barth and other
relevant isles occupied by the French before the integration "en masse" of
African-born slaves and consequently that:

IMAGE 01h2LexDiffFin16.gif
IMAGE 01h2LexDiffFin16.gif

The Saint-Barth continuum of today captures the gradualness of diffusion at the
very time when, for socio-historical reason, the diffusion of the changes such as
the deletion of overt agreement markers froze in time.

The residual question would be at what time agreement deletion occurred:Was it
already incipient in koine varieties that were imported or did it develop in situ?
In this respect, it is interesting to note that the White population of Martinique
and Saint Domingue (Haiti) at the close of the 17th century had an unusually high
input of immigrants hailing presumably from outside the Gallo-Romance speech area,
39% and 58% respectively (cf. Wittmann 1995:292). The ratio might have even been
higher among the "gens de bois" who more than anybody else maintained relations with
the local Indians.



IMAGE 01h2LexDiffFin41.gif

After all, I agree with Wittmann, Mufwene, DeGraff and others that there are no
specific features that turn up in creoles but not elsewhere, but only large clusters
of such features.

In any case, I would like to cite what Ernesto d'Andrade and Alain Kihm said in a
splendid article in the recent book "Crioulos de base Portuguesa", namely that
regardless of whether non-creoles do what creoles do, "the significant observation

IMAGE 01h2LexDiffFin01.gif


Printed for Henri Wittmann <>

IMAGE 01h2LexDiffFin14.gif



is that, every time a language recordably became a creole, the resulting grammar
shows a less extensive PF process [=discrepancy between underlying semantics and
phonetic form] than do its source language [=lexifier] and its putative substrate
languages" (Andrade & Kihm 2000a:11).

This looks a bit contradictory to me. If on one hand you agree there is no creole
prototype but on the other you agree also that creoles can do some things regardless
of what non-creoles do, where is the loophole permitting to falsifiy whatever you're
agreeing on? In which way can opinions claiming that a particular language
recordable, let's say Haitian Creole, became a creole, be falsified?

IMAGE 01h2LexDiffFin23.gif

This is what interests me the most, more than substrate influences or, for that
matter the issue of whether proto-Haitian was absolutely affixless or not (for the
record: many a creole, possibly even most, has one or two affixes, so early Haitian
may well have had that too).

Unfortunately, that interest comes posterior to "Out of Africa".

IMAGE 01h2LexDiffFin04.gif
IMAGE 01h2LexDiffFin16.gif

Any language, even the most isolating one, would have to have inflection,
fusional, clitic or isolating, in order to qualify as a natural language. Any
non-lexical "functional" item is inflective by definition.

No one has ever contested that all languages have grammatical morphemes. What I
would contest is what DeGraff seems to be saying, viz that a language must have
bound morphemes in order to classify as a human language. As both you and he (and
presumably everybody else on the list) knows, your use of the term "morphology"
isn't the only one, and probably not even the most common.

You extrapolated that quotation out of a footnote on McWhorter, nothing to do with
CONTEXT" in order to be able to tar me with the same brush as someone I don't even
know or to be able to tell "everybody else on the list" that I have uncommun tastes
in morphology? As "everybody else on the list" also knows, you're simply angry at
me, and quite understably so, for ruining your hypotheses. So, no one "on the list"
will chide me if I don't respond in detail to insinuations of that kind.

A final question I'd like to answer without being really asked to is why I style the
preverbal /i-te-àpre-à-/ in an example such as:

(42)amwen ch-te-àpre-à-manjé à-pòm-la

as clitics rather than affixes. The answer for right now must be that promoting
clitics to affixe status in the Magoua data would entail a similar decision in
corresponding Haitian Creole data:

(42)bm-t-ap-manjé mango-a

which would make prototypers unhappy (I'm referring here to McWhorter, not to
Parkvall who confessed not to be a prototyper) besides being irrelevant to a debate
on the comparative origins of koine and creole varieties of French.

IMAGE 01h2LexDiffFin04.gif

Although the following comment by Wittmann is not directed towards me personally...

IMAGE 01h2LexDiffFin16.gif

Nativized pidgins, inappropriately called creoles, borrow their vocabulary to a
greater extent from the superstrate language though the resyntactification cannot
be shown to derive clearly in most cases from any substrate input.

... I would like to add, given that the discussion above (as well as my recent book)
dealt with substrate influences that I agree fully with Wittmann, with the exception
that I see nothing inappropriate in calling nativised pidgins "creoles".

As long as you accept that what we call conveniently "French Creoles" are not
nativized pidgins "out of Africa".

IMAGE 01h2LexDiffFin16.gif

While Wittmann sees me and others as claiming that "varieties of French spoken by
descendants of immigrants from Africa developed FOR NO APPARENT REASON in a manner
not similar to varieties of French spoken by immigrants from elsewhere" (emphasis

IMAGE 01h2LexDiffFin01.gif


Printed for Henri Wittmann <>

[made with GoClick]