IMAGE 01h2LexDiffFin14.gif

4

01h2LexDiffFin

I would like to propose that GALERIE is probably a French loanword in Haitian. In
which case /e/ for pre-tonic schwa would be the regular reflex for French LOANS in
Haitian, and /i/ for INHERITED lexical items.

GALERIE is the general term for what it seems to be in HC not only in HC, but in ALL
varieties of Colonial French of the Americas, corresponding to VARANGUE in ALL the
Indian Ocean varieties. First attestation of <galerie> in the Americas probably is
Journal des Jésuites p. 236 (ap. Clapin 1894:165):"Le 22, durant qu'on tenoit
conseil sur la galerie du fort ..., la dite galerie se rompit par le milieu."
Surface forms in HC are galeri, galri, and gayri, all three forms occurring also in
Magoua as galuri, galri, gayri.

As to the more specific question of e/i (and similar variations) in the 17th koine,
we have to distinguish between (I) e/i resulting from underlying segments and (II)
e/i resulting from schwa-insertion.

(I) If the situation in Magoua (Wittmann 1996) and related dialects is any
indication, let's assume that the 17th koine had, not considering for the moment low
or back vowels, six front vowels:two long unrounded /é e/, one long rounded /eú/,
two short unrounded /i è/, and one short rounded /u/ (if there was also a long /eu/
and a short /eù/ is a matter for debate). Let's also assume that the short front
vowels are the reduced (lax) counterparts of the long ones and that a rule of vowel
reduction operates in unstressed syllables (the rule also operates with word-final
open syllables as we'll see later). The result would be that underlying é/i such as
in /érit/ "inherit" : /irit/ "irritate" would invariably come out as [IRIt] which is
actually the case in ALL varieties of Québécois (aka the Laurentian koine varieties
of Colonial French). Yet, the interpretation of this case of homophony as [IRIt],
[eRet], [Iret] or [eRIt] is the matter of a long-standing debate (see indicatively
McLaughlin 1982, Picard 1987, Wittmann 1996), though the contributors to that debate
are mostly native speakers of the language under debate. (In naive writing, <é> for
unstressed <i> such as in <ménuit> for <minuit> seems to be the rule in most if not
all glossaries.)

(II) As for schwa-insertion, the situation is even more muddled. Though phonologists
are under no obligation to tell us what the exact phonetic nature of the inserted
schwa might be, phoneticians can't agree on any specific color for schwa phenomena in
Québécois. In naive transliteration, <u> for schwa (such as in <chuville, chumin,
chuminée> for <cheville, chemin, cheminée>) seems to be the most common in all
positions not involving a follwing <r>, though <i> is not rare (as in <chimise> for
<chemise>; cf. here also the IO varieties). As for <re->, the glossaries give
interchangeably <er-, eur-, ar->;<mercredi> is variously rendered as <me(n)kerdi,
me(n)keurdi> or <mécredi, mécridi>.

If we can extrapolate from a dilemma afflicting the "pure wool" Québécois, whether
trained in linguistics or not, interpreting the e/i variation as symptomatic for
BORROWED vs INHERITED seems to me amateurish, no offense meant, coming from somebody
who seems to be a pure wool French-Canadian on the payroll (or aspiring to be) of
some departement of French or linguistics at a French-Candian university.

IMAGE 01h2LexDiffFin05.gif

Stephane Goyette writes further (07/02/01):

> As early as 1331 (!) the word "galerie" was [...]

Alain Rey is normally highly reliable, so this certainly casts sufficient doubt for
me to withdraw my earlier remark on "Galerie".

In this struggle with UD, you have "withdrawn" a lot of times. Maybe you should
check things out more carefully before throwing them around.

IMAGE 01h2LexDiffFin16.gif

[...]A third example of pre-tonic schwa yielding /i/ is /tini/, from TENIR. But I
don't see what Professor DeGraff's "counter-examples" prove, if anything: I would
maintain that they indeed are borrowings.

If three examples with "pre-tonic schwa yielding /i/", vini, tini, sila, is all you
have to go on for demonstrating that only items made up according to your rules
qualify as native vocabulary, it would amount to postulating that HC has almost no

IMAGE 01h2LexDiffFin01.gif

4

Printed for Henri Wittmann <hwittman@cgocable.ca>

IMAGE 01h2LexDiffFin01.gif

5

01h2LexDiffFin

native lexicon to speak of.

IMAGE 01h2LexDiffFin19.gif

I may be wrong. [...]

Can you afford to? Would you want a surgeon on the job to gamble along such
assumptions?

IMAGE 01h2LexDiffFin06.gif

[...]Are Haitians "magically" unaffected by the principle of regularity of sound
change?

You might want to consider [e] to be the regular reflex of pre-tonic schwa with [i]
being the exception.

IMAGE 01h2LexDiffFin21.gif

>resevwa vs. *resivwa (cf. French recevoir `receive')

Why is it /resevwa/ and not */resevwè/ or */resewè/? French VOIR "to see" appears
in Haitian Creole as /wè/.

Now, historically, the diphthong spelled (OI) in French shifted from /we/ to /wa/
during the last decades of the eighteenth century. If Haitian Creole had inherited
VOIR (with /we/) but borrowed RECEVOIR, at a later date (when /we/ had become /wa/),
this would fit the facts nicely. And confirm that pre-tonic schwa IN LOANWORDS
yields /e/.

The /wé/ shifted to /wa/ during the last decades of the 18th century IN STANDARD
FRENCH. In the Parisian koine that was exported the change to /wa/ had started to
compete with /we/ much earlier. This is why speech varieties of the earliest
settlements of the 1st half of the 17th century (such as the French spoken between
Québec and Montreal) have predominantly /we/ whereas 18th century settlements (such
as Mauritius) have generalized /wa/. Even Québécois has basic lexical items with
/wa/ that cannot be accounted for by "borrowing":<doigt> "finger", <oie> "goose",
<toit> "roof", for instance, have /wà/ with a front low vowel; <bois> "wood", <poids>
"weight", <trois> "three", on the other hand, have /wa/ with a back low vowel;
whereas <boit> "drinks", <doit> "must", <toi> "you" have "regular" /wé/. If /wé/
were to be the only option, "roof", "three" and "you" would have to be pronounced
/twé/ which, as a "native speaker", you know to be untrue except for "you".

IMAGE 01h2LexDiffFin22.gif

>jedi vs. *jidi (cf. French jeudi `Thursday')

This unstressed vowel (EU) wasn't identical to schwa in the seventeenth century
(Modern Quebec French speakers, including this writer, still keep JEUDI and JE DIS
quite distinct).

You have been digging your own grave in this respect by not distinguishing in any
theoretically meaningful way between root-based segmental phenomena and phonetic
facts resulting from insertion. Two of your three examples for "pre-tonic schwa
yielding /i/", vini and tini, can be shown to have underlying /i/ on the basis of
evidence from the varieties where vini and tini syncopate to vin and tin. The 17th
century koine most certainly supported forms such as <vinir> and <tinir> realigned on
the model of <finir> competing with <viendre> and <tiendre>/<quiendre>. Highfield, a
source you claim to have worked with, has /vèni/ with a front [E], not */veni/ with a
schwa as one would expect to find for a non-creole variety of French if your
predictions were well founded. If Highfield had wanted to insert a schwa, he would
have done so as he did in many other instances such as the transcription of
ethymological <re-> as /r/ + schwa.

IMAGE 01h2LexDiffFin23.gif

>rejè vs. *rijè (cf. French rejet `reject')

Interesting example. Why wasn't final /t/ preserved, as it is in most Haitian words?
[...]

Even in Québécois, this /-t/ is unevenly preserved.

IMAGE 01h2LexDiffFin23.gif

Since /e/ clearly *does* correspond to French schwa in loanwords, as I hope to have
shown above, to assume that *all* /e/-schwa correspondances between French and
Haitian are due to French loanwords is the simplest hypothesis, and should be one's
working hypothesis until evidence to the contrary is adduced.

IMAGE 01h2LexDiffFin01.gif

5

Printed for Henri Wittmann <hwittman@cgocable.ca>

IMAGE 01h2LexDiffFin01.gif

6

01h2LexDiffFin

The simplest hypothesis is to see the e-schwa/i-schwa alternation as the outcome of
competing changes. Ever heard of lexical diffusion? Wang (1969) came up with the
hypothesis that a phonological change, as it diffuses across the lexicon, may not
reach all the lexical items to which it may be applicable. If a competing change
intersects at the right time and in the right place, residue may result. You might
want to read more on these and other issues before attempting to explain phenomena
others before you have already dealt with extensively (see for example Labov
1994:421-543 for a state-of-the-art account of regular sound change vs lexical
diffusion).

IMAGE 01h2LexDiffFin10.gif

Bethanie Morrissey (commenting on UD) wrote (08/02/01):

"Erosion" by definition means GRADUAL. You are confusing "erode" with "amputate."

> Interestingly, the erosion of grammatical gender in the history of
> English would turn English into a "creole".

If it happened abruptly due to abnormal transmission, it sure might.

"Gradual" doesn't mean "slow". I would refer you to the same readings. As Wang
summed it up so nicely in 1969,change is triggered abruptly but may diffuse
gradually across the lexicon. As it is, the span of time required is largely
determined by non-linguistic factors, not by a priori parameters capable of
distinguising between normal and abnormal transmission. Weinreich called this the
congruence of linguistic and non-linguistic factors. Later on, people in
sociolinguistics started to talk about co-variation (of linguistic and social
variables) and even the linguistic market. Read about it, it's fun.

II. ON CONVERGENCE AND BEYOND:"CREOLISMS" IN KOINE GENESIS

IMAGE 01h2LexDiffFin22.gif

Bethanie Morrissey wrote (31/01/01):

[...] I said from the very beginning that if we didn't know what time frame these
data are from, they aren't convincing. And THAT is what started this whole debate.
Remember the convergence issue?

I do remember. I also remember her reply to my starting off the debate on
"Convergence:Nothing to do with creolization":

IMAGE 01h2LexDiffFin19.gif

Before I continue, I must know if Michel DeGraff agrees with your input.

which showed that she is more concerned with debunking Michel DeGraff than with
hearing about convergence from linguists (assuming that most debaters on this list,
with some notable exceptions, are creolists rather than linguists). I'm singling out
here the same creolist who in debunking DeGraff wrote (17/08/99):

IMAGE 01h2LexDiffFin06.gif

Historical linguistics restricts its studies to languages which will fit on the
family tree, deeming creoles unworthy.

As it is, historical linguistics is an "area" of linguistics resulting from the
convergence of three 19th century trends:(a) Stammbaumtheorie (the only theorie of
HL she admittedly knows anything about),Wellentheorie (the trend of HL Schuchardt
contributed to, ringing any bell?), and Sprachtypologie (classification of languages
without reference to their genealogical "family tree" classification). The latter
trend got Sapir to write on "drift" (in 1921) and ultimately (from 1931 on)
Trubetzkoy and Jakobson on "Sprachbünde", a concept Weinreich translated as
"congergence areas" such as we traditionnally knew about in the Balkans and the
Indian subcontinent. "Convergence" in this context was meant to imply something that
Ronald Kephart (19/01/01) aptly analogizes as *gene flow*, a kind of bahavorial
mechanism based on collective "mimetism" that over time got the Albanians, the
Bulgarians, the Romanians, and the Turks to share the syntactic feature of noun-based
definite direct object marking (DDOM) or which can account for the diffusion of
benefactive serialization in West-Africa (Hyman 1975). It wasn't meant to account
for or to theoretically pave the way for large-scale conflations or massive calquing,
let alone to open a theoretical loop-hole for the emergency of "New Languages". It

IMAGE 01h2LexDiffFin01.gif

6

Printed for Henri Wittmann <hwittman@cgocable.ca>

[made with GoClick]