IMAGE 01h2LexDiffFin01.gif



(21)aj'ai apporté un livre pour Marie(Standard French)
(21)bj'ai apporté un livre à Marie

(22)a(mouén) j-vyen l-mné l-liv pou Mari(Magoua)
(22)ba(mouén) j-vyen l-mné l-liv à Mari
(22)bb(mouén) j-vyen i-mné Mari son liv

[In Magoua, /j-vyen/ is structurally ambiguous:It can reflect underlying <je
viens>, <j'ai vient/venu> or <je suis vient/venu>.]

(23)amè ná àgbàlê Kòfí(Gbe)
"I gave [= passed on] book Koffi"
(23)bmè tsóàgbàlê ná Kòfí
"I gave [= gave as a gift] book to Koffi"

A careful confrontation of the data offers the following riddle which should
embarrass both substratists and superstratists alike:Why does the distribution of
ba/pou in the ba-varieties of creole French run counter the models of both Gbe (and
other African languages) and French (including Magoua, the dialects related to Magoua
and the non-ba varieties of creole French)? If the serial /ba/ had the same
benefactive properties as the /ná/ of Gbe and if the preposition /pou/ had retained
the lative properties of <pour> in French, a nice case of "convergence" would have
resulted from a distribution maintaining congruent substrate and superstrate

The case for French offers the following vantage points:The verb <bailler> (the
short stem being usually <ba->, future tense:<barra>) has a non-benefactive lative
meaning, "carry, take hold off, pass on, hand over", in all the varieties of French
where it subsists or ever existed. In Picard, it supports lative serialization
(Fournier 1994). I have been able to show that, (a) though no Picard immigration of
import into the New World ever took place and (b) no trace of <bailler> subsists in
any koine variety of French, Picard was a crucial input to the koine that was
eventually exported (Wittmann 1995). As it is, a koine related transmission model is
the only conceivable link between Picard and the ba-varieties of creole French. More
recently, I was able to substantiate an occurrence of serial /bay/ in the non-koine
dialect documented to have subsisted into second half of the 19th century by Nisard
(1872, 1876):

(24)baille-ly belle, la queue ly pu(Nisard 1876:17)
"At it truly, his tail stinks"

[Apparently, when examining game at the market place, people who knew what they were
doing would sniff the ware "at the root of the tail". The dialect is, as Nisard
points out himself, heavily picardized.]

However, this doesn't explain why, though the lative interpretation subsisted with
the serial, the verb itself took on in ba-creole French benefactive interpretation
displacing totally the original <donner> "to give" that competes with <bailler>
everywhere else.

Against substrate induction, it must ultimately be considered that true benefactive
serialization, in all the French-derived pidgins of West-Africa, is invariably based
on French <donner> alternating synonymously with prepositional <à>, never on any
other lexical input, and that the lative alternate is correspondingly based on

On the other hand, the explanation for the distribution of ba/pou as it stands
remains sufficiently anomalous to render a substrate contribution exegesis virtually
indispensable. I have offered a viable alternative by suggesting that the
distribution of ba/pou reflects a case of conflation with karipuna ba/pu (Wittmann
1987a, 1994).

[Karipuna is a relexified version of a now extinct Arawakan language. It has overtly
typical Arawakan syntax with a vocabulary largely derived from Karina. The latter is
a Kariban language with a heavy lexical adstrate from Tupi. Karipuna was the
dominant vehicular language in all the territories where the ba-varieties of creole

IMAGE 01h2LexDiffFin01.gif


Printed for Henri Wittmann <>

IMAGE 01h2LexDiffFin01.gif



French are now spoken competing with Arawakan languages in the isles, Arawakan and
Kariban languages in the Guianas, including Karina itself. The only surving variety
of Karipuna today is Garifuna, the language of the black population of Saint-Vincent
the British resettled in Central America in the early 19th century. Though the Gbe
input was very strong in the demographic make-up, Garifuna is not a creole by any
means and demonstrates that Africans are able to learn a new language even when they
vastly outnumber the superstrate speakers. Surviving varieties of Karipuna spoken by
Indians on Dominica, Trinidad in northern Brazil died out during the first half of
the 20th century. Because of the heavy Tupi adstrate, varieties of Karipuna or
Karina have sometimes been mistaken for Tupi derived languages.]

[Conflations are "contact friendly" accidental similarities between two languages, an
early example in the literature being the prepositionpaof Russian and Norwegian.
Conflations such as these are said to have helped along the emergence of Russenorsk.
Conflations are not direct inputs corresponding to borrowings or calques, they are
contact-induced cues that a source language might have to offer in helping along
adaptive changes or the diffusion of on-going changes in the target language. As
such, they conform to Jakobson's (1938) requirement that substratal models must not
be incompatible with the latent tendencies of the target language.]

Though both /ba/ and /pu/ have lative interpretation in Karipuna, /pu/ is a
Karina-derived true preposition homophonous with French-derived /pou/, whereas /ba/
is an Arawakan-derived true serial alternating in Karipuna with a true verb /ba/ "go"
which happened to be homophonous with a French-derived verb root /ba-/ "to give"
which had alternating lative-benefactive interpretations. [The /ba/ of Karipuna has
also functions of aspect-tense and as a complementizer.]

The case for conflation can be argued with other instances of substrate input from
Karipuna attributable to accidental similarites between Karipuna and French:


(a)ke "with, and" (HC ake, Karipuna Creole ke):Karipuna ke "with, and".

(252)ka "to become, to be able, could" (French <qu'à>, replaces HC /ap/
[progressive] in all ba-varieties except HC):French <qu'à> (cf. Magoua /ch-ka fer
sa/ "I just have to do that"), Karipuna k- [attributive] + -a- verbalizing prefix, or
-ka [stative, durative] (cf. Palikour -ka/-ki [+passive, -perfective] as in /ri-hepka
tamak-ka duwwe/ "he-face painted-is red", Karipuna Creole /so fas ka pentire rouj/
"his face is painted red").

(253)kaba "finished, after, already":Karipuna k-a- (as in 252) -ba
[imperfective] (-ba is past imperfective as an aspect, future as a tense), originally
conflating with Spanish <acabar> "complete".

(254)ke [future] (in all ba-varieties except HC, from an earlier k(a)- (as in 252)
+ ale/ay):Karipuna k-a- (as in 252) + a conflating calque based on Karipuna ba
"aller" ou -ba [future].

(255)ki "which", ki-sa "what" (in Sylvain 1936:71 also for "who"), sa "what, that,
it is":Karipuna ka/ki (a phonological rule determining the distribution) "what,
which, who", ka sa "what" (also employed to form phrases meaning why).


la DET:Karipuna l(i-k)a [demontrative] (meaning as in French "ce ... là").

(257)li, l, i "he, she", competing with "he/his, she/her":Karipuna l-, -i "he,
his", t-, -u "she, her".


pa- [negative]:Karipuna -pa [negative] (competes with privative ma- as in


t(e) [past]:Karipuna -t(i)- [aorist].

(25A)pou/ba:(This is the alternation we already talked about which came here in
the list.)

(25B)(v)a [future] (from earlier "go" in French, competes with ke of 254):
Conflating calque based on Karipuna ba "to go", -ba [future]. Conflating calquing
"helping along" in the ba-area is not inconsistent with the fact that forms based on

IMAGE 01h2LexDiffFin01.gif


Printed for Henri Wittmann <>

IMAGE 01h2LexDiffFin01.gif



/va/ generalized elsewhere.

(25C)yo/ye "they, their, -s" PRO-POSS-PL:Karipuna -yen, -yan, -yón, same meaning
(I used here in 25C transcription conventions borrowed from Magoua).

As for the lexicon, a number of items preserve privative /ma-/ (conflating with
French-derived <mal>) or the agglutinated feminine article /-t/ (conflating with the
derivational suffix <-te> from French):

(26)mabi "boisson (sp.)", mabuya "mauvais génie, lézard (sp.), se remuer", mafya
"esprit malveillant, canaille", maho "arbre (sp.), fibre (sp)", makawa "rhumatisme",
makwa "sans élégance", makwali "mal bâti", manarè "crible (de farine de manioc)",
manyòk "manioc", mapu "kapokier", marasa "jumeaux", maraka "hochet (de calebasse)",
marengwen "moustique", maribouja "fruit de la passion", maripa "sans harnachement",
masisi "sans testicules, homosexuel", masuru "sans parents", matabi "presse à
manioc", matete "besoin, manque", matuni "bigorneau", mayipuri "tapir", matutu
"plat/riz de crabe", maveve "plante (sp), remède (sp.)", mawali "petit poisson
(sp.)", mayayuka "moquerie, jeu de dérision", mayi "maïs", mazora "sans dents,

(27)kannò-t "barque, canot", karè-t "tortue de mer", kayimi-t "caïmite", maku-t
"sac en paille", pipiri-t "petit oiseau (sp.)", ravè-t "blatte, cancrelat", zandoli-t
"lézard (sp.)"

As a general rule, it can be said that the Karipuna-derived lexical input is the most
important one after French in the lexical makeup of the ba-dialects. Karipuna has
also nasal vowels just like in the ba-varieties of creole French.

Some final thoughts on serialization:

Though one must admit that Parkvall has amassed a wealth of information as far as the
non-French atlantic creoles and their suspected substratal roots are concerned, I
can't help feeling that his perception of verb serialization in the relevant African
languages is somewhat daintily perverted with ditherings of unconscious albocentrism.
Parkvall's interest in verb serialization focuses on the spectacular aspects of the
phenomenon which happen to be also the most superficial ones. Productive verb
serialization is a strategy to resolve lexico-semantic complexities syntactically as
a string of semantic primitives. With verbs, you would expect to find semantically
complex action concepts such as BRING TO in the syntactic context of (28a)


somebody+tense something BRING-TO somewhere

(which in a European language would ordinarily be expressed by a single verb plus a
non-nominative case marker such as a preposition) resolved consistently as a series
of verbs such as in (28b)



The occasionally arising apparent categorial fuzziness is purely incidental to the
process and is obvious to the minds of European observers and literate bilinguals
only. A consistently serializing language such as Yoruba has no such things as
prepositions or postpositions. Educated bilingual Yorubas of today "tend to usefún
rather indiscriminately to translate 'for'" (Rowlands 1969:84), no such behavior is
to be expected from enslaved Yorubas of the 17th century. Similarly, Gbe which has
postpositions and productive verb serialization, had in the times of Westermann just
three "adverbs" grammaticalized out of verbal serials:"there", "and then", "ready",
but no "prepositions" (1939:48). Lafage who had the occasion to examin the
grammatical status of serial GIVE more recently had effectively this to say
(1976:413-14, my translation):"...has a tendency to no longer requiring verbal
modalities and is about to become a sort of preposition 'à la French' equivalent of
"to, for", Gbe having postpositions normally. Is this 'neutralization' of the verb
the product of interference from French? The phenomenon seems to be very recent."
This looks rather like evidence to the effect that Africans in recent times have
imported into their native languages grammatical concepts remindful of prepositions
and adverbs, not that Africans in the 17th century exported serializing stategies in
European garb.

IMAGE 01h2LexDiffFin01.gif


Printed for Henri Wittmann <>

[made with GoClick]