01h2LexDiffFin

01h2LexDiffFin

LEXICAL DIFFUSION AND THE GLOTTOGENETICS OF CREOLE FRENCH

Henri Wittmann
Professeur émérite de linguistique
Directeur, Presses Universitaires de Trois-Rivières
<hwittman@cgocable.ca>

INTRODUCTION

There are six parts and nine appendixes to this article: I. On lexical diffusion;
II. On convergence and beyond: "Creolisms" in koine genesis; III. The frills of
French in the genesis of creole French; IV. Serialization in French; V. Conclusions;
VI. References; Appendix 1: Statement on convergence; Appendix 2: re- affixation in
French; Appendix 3: On gradualness; Appendix 4: On theoretical bias; Appendix 5:
Replying to Stephane Goyette I; Appendix 6: Replying to Stephane Goyette II; Appendix
7: The "sontaient" syndrome; Appendix 8: Replying to Stephane Goyette III; Appendix
9: Reply to Parkvall (26/02/01). Appendix 9 is divided as follows: 9.0. On dishonesty and
unscholarly behavior; 9.1. The "sontaient" syndrome; 9.2 Noun class system death and determiner system renewal; 9.3. The
syntax of comparative constructions; 9.4. More on verb serialization; 9.5. Bimorphemic interrogatives; 9.6. The Saint-Barth
"enigma"; 9.7. Varia; 9.8. The rest of Out of Africa; 9.9. Concluding on Parkvall.

Parts I-VI are bits and pieces on the subject of "Lexical diffusion and the
glottogenetics of creole French" which were originally posted on CreoLIST as a series
of rejoinders to contributions to a debate centering around data from Haitian Creole
as evidence for abnormal transmission (DeGraff 2001), with contributions by Hildo de
Couto, Michel DeGraff, Emmanuel Faure, Stephane Goyette, Ronald Kephart, Bethanie
Morrissey, Salikoko S. Mufwene, Mikael Parkvall, myself (18/01/01, 04/02/01) and
others who for reasons of space I don't dare to quote. Appendixes I-II are
statements on convergence and re- affixing in French I posted prior to posting part
I, Appendixes III ff. are my replies to replies subsequent to the posting of part VI.

A mime-compliant handout in PDF format for examples 1-29 is available at:

< http://cafe.rapidus.net/hwittman/01h1-glottoV2.pdf >

and alternately at:

< http://homepage.mac.com/noula/01h1-glottoV2.pdf >

A mime-compliant handout in PDF format for examples 30-47 in the appendixes is
available at:

< http://cafe.rapidus.net/hwittman/01h1-glotto9.pdf >

and alternately at:

< http://homepage.mac.com/noula/01h1-glotto9.pdf >

I'm not taking up any of DeGraff's arguments directly. The perspective taken here is
comparative. Footnotes appear within brackets in the text. There has been no
attempt at postediting or revising anything. More information on "CreoLIST" is
available at:

< http://www.ling.su.se/Creole/ >

For full online access of all contributions the debate, see the CreoLIST archives at:

http://listserv.linguistlist.org/archives/creolist.html

The contributions appearing here are archived at the follwing URLs:

http://listserv.linguistlist.org/cgi-bin/wa?A2=ind0102&L=creolist&P=R10155 (P1)
http://listserv.linguistlist.org/cgi-bin/wa?A2=ind0102&L=creolist&P=R10609 (P2)
http://listserv.linguistlist.org/cgi-bin/wa?A2=ind0102&L=creolist&P=R10921 (P3)

Printed for Henri Wittmann <hwittman@cgocable.ca>

01h2LexDiffFin

I. ON LEXICAL DIFFUSION

Stephane Goyette writes (05/02/01):

I am grateful to professor Wittmann for his arguments on the presence of (E)R- in
seventeenth-century French. I accept that we are dealing with schwa insertion
rather than with metathesis, although synchronically such RE-/ER- alternation is
identical to a process of metathesis, so I trust nobody will mind that I still refer
to it as "metathesis" (see Flutre's first volume (phonological evolution) of DU
MOYEN PICARD AU PICARD MODERNE, where a similar point is made). Similar forms are
documented by Highfield for Saint-Thomas French, incidentally.

At a stage where in historical linguistics cases of so-called "metathesis" are being
consistently reinterpreted as performance errors (tongue slips, spoonerisms, etc),
that is to say as phenomena which by their very nature cannot be "regular" (which in
turn explains why there can't be a phonological rule of metathesis "synchronically"
productive), I would strongly object to productive schwa-insertion being equated to
"a process of metathesis". Flutre was no phonologist by any means and had he
bothered to consult on this matter, let's say, his good friend Martinet, an authority
on such questions in 1977, he'd have been able to distinghish segments with
phonological status from non-segmental material (such as schwa-insertion,
affrication, etc.) with no such status. As it is, only segments can permute.

In varieties where postvocalic /r/ deletes in unstressed positions (such as in
Magoua), you get eùkulé for <reculer>.

Which fits Haitian /ekile/ perfectly. I should perhaps explain [...] why, on the
basis of my hypothesis that Haitian Creole had gone through an affixless stage, I
*expected* to find a form /ekile/.

Before you go any further, I must assume at his point that you consider HC to have
"gone through an affixless stage" because the rule of schwa-insertion before the /r-/
seems to be unproductive today whereas the productivity of the rule in Magoua implies
that no such stage is to be expected in Magoua (or in any other dialect with
productive schwa-insertion).

Haitian Creole has a prefix RE-, always attested in this phonological form according
to Fattier. PROBLEM: why do we *not* find forms of the *prefix* such as E(R)- or
A(R)- (with later deletion of the rhotic)? The comparative evidence certainly
suggests such a form was present in the variety of French which gave birth to
Haitian.

Now, let us suppose that Haitian has indeed *borrowed* this prefixed
(unmetathesized) RE- from Standard French, and was originally a language lacking
bound morphemes (like Ndjuka). Where would one expect to find attestations of
metathesized RE-? Why, in reflexes of French lexical items which have initial RE-
*but which are monomorphemic*. Furthermore, one would expect to find it in basic
lexical items, which are likelier to be inherited (rather than borrowed) elements.

RECULER fit the bill perfectly: it is semantically basic. No verb *CULER exists.
Imagine how pleased I was to indeed find attestations of /ekile/ in Fattier's Atlas.

Printed for Henri Wittmann <hwittman@cgocable.ca>

01h2LexDiffFin

At this point, I assume that such attestations in a non-creole dialect would NOT
please you, right?

This explains why the *prefix* RE- does not have any *E- or *A- allomorph, but
reflexes of French RECULER do. Professor DeGraff should note that Réunionnais,Magoua
(and other Laurentian French varieties), Picard and Western langue d'oïl varieties
show an allomorph E(R)-, A(R) of RE- *whether it is a prefix or not*. Why isn't this
the case in Haitian?

The scenario sketched above, whereby Haitian was originally an affixless language,
explains it nicely.

Not at all. You just brought in yourself evidence from Saint-Thomas French, an
expatriate dialect of the koine variety spoken in the western part of
Saint-Barthelemy which I told you to have unproductive schwa-insertion in front of
/r-/ (such as in /Arkul arkul/ "Hercules step back"). Even if you've got only
Highfield to rely on for this, it's obvious from the glossary (pp. 228-347) that
/re-/ is the norm and /a(r)-/ the exception, just like in Haitian Creole.

If you think Saint-Barth is too close to the creole area, then let's take the koine
variety of Bonaventure County which is enclaved between the non-creole non-koine
Acadian dialect area and the non-creole koine varieties of Laurentian (we already
talked about), both these surrounding areas maintaining productive schwa-insertion.
In Bonaventure French, schwa-insertion in front of /r-/ is just as exceptionnal as in
Saint-Barth French and Haitian Creole.

If you were thinking of dealing the Ugly DeGraff a final blow with this, you were
definitely taking a shot in the dark regardless of whether re-/er- alternations are
explained in terms of metathesis or schwa-insertion.

Stephane Goyette writes elsewhere (31/01/01):

[...] *For the sake of argument*, let us suppose I agree that (CONTRA all the
comparative evidence) the variety of French ancestral to Haitian did not metathesize
RE-. The phonological evidence would still not allow for Haitian /re/ to be
inherited from French.

Why? Forms such as Haitian /vini/ or /sila/ (from French VENIR and CELA) make it
plain that (French) pre-tonic schwa yields /i/ in Haitian Creole. I would thus
expect the form of the prefix to be */ri/, not /re/.

Michel DeGraff's earlier comment (01/02/01) quoted by Goyette:

Even as a non-phonologist, I am able to produce a straightforward preliminary list
of counterexamples of French pre-tonic schwa has yield /e/ not /i/. I capitalize
the relevant vowels (the /e/ and the schwa in the _attested_ Haitian and French
forms, respectively; and the /i/ in the _unattested_ Haitian form---capitals
aside, all forms are written in the standard spellings):

chEmiz vs. *chImiz (cf. French chemIse)
dEmen vs. * dImen (cf. French dEmain)
galEri vs. *galIri (cf. French gallErie)
jEnou vs. *jInou (cf. French gEnou)
lEve vs. *live (cf. French lEver)
mEnwat vs. *mInwat (cf. French mEnuet)
resEvwa vs. *resIvwa (cf. French recEvoir)

Stephane Goyette's reply (05/02/01):

So which is it? Does pre-tonic schwa yield /i/ or /e/?

> galEri vs. *galIri (cf. French gallErie)

French GALERIE (spelled with a single L, by the way) is only attested in its modern
sense in the 1780's. Earlier, it was a specialized architectural term referring to
the outer part of a monastery.

Printed for Henri Wittmann <hwittman@cgocable.ca>