On Mon, 28 Jan 2002 23:38:49 -0000, Lars Henrik Mathiesen <[log in to unmask]>

>> Date:         Mon, 28 Jan 2002 15:49:04 -0500
>> From: Vasiliy Chernov <[log in to unmask]>
>> On Mon, 28 Jan 2002 10:43:19 -0000, Lars Henrik Mathiesen
<[log in to unmask]>
>> wrote:
>> >French /d2/: not too weird, but I'm too lazy to look up how it came to
>> >have a front rounded vowel.
>> duo:s > do:s > dous > d2ys > d2:x > d2h > d2: > d2
>Thanks --- I guess that's fronting caused by the /s/.

No. Actually, I should have given more details.

duo:s > do(:)s (contraction + loss of quantity distinctions) - in (most
dialects of) VL;

do(:)s > do:s (new phonemization of quantities in stressed syllables
depending on the original syllable structure; /o:/ now opposed to both
/o/ and /O:/) - in Gallo-Romance;

do:s > dous (a diphthongization parallel to /e:/ > /ei/ and emphasizing
the diference from /O:/ > /uO/) - in late Gallo-Romance or very early Old

dous > dys (graphically {ou} > {eu}; a change parallel to /u/ > /y/ {u},
and /uO/ > /y/, graphically {uo} > {ue}) - in Central French ( = Francien)
by (the end of?) 11th century.

dys > d(:)s (monophthongization, perhaps associated with new loss of
quantities) in later Old French (which is hard to date precisely, since the
spelling did not change; but confusion of {eu} and {ue} seems to happen as
early as in 13th century, at least occasionally).

>By the way, is
>that x an X-SAMPA /x/? I thought that French just dropped consonants
>in absolute final position, no bother with intermediate lenitions...

I don't know for the other consonants, but /s/ > /x/ (syllable-finally)
seems to be a well established change. An evidence I can recall is based
on Middle High German transcriptions of (late) Old French (there are quite
a few transcribed French phrases e. g. in minnesingers' poems). For
example, with _est_ written {echt} (I don't remember the examples for
word-final {ch}, but it seems there were some). I sort of recall there
were also other evidences which I forget.

Curiously, in the history of French, phonemic  vowel quantities then
re-apperared once more (which involved changes like /Vx/ > /V:/, /@V/ >
/V:/, /au/ > /o:/) and finally were lost again (yielding new quality
distinctions like /a/ vs. /A(:)/, /O/ vs. /o(:)/, etc.). That is, loss
of quantities happened thrice in less than two millennia!

>> >Armenian /Erku/: /d/ and /r/ can both be realized as a flap [4], /w/
>> >often has some velar constriction, and then an epenthetic vowel:
>> >
>> >    */dwo/ > [4wo] > [4ku] > /rku/ > /Erku/
>> Or perhaps dw- > dg_w- > rg_w- > rk_w- > ...
>Indeed, a voiced labiovelar is more plausible as the intermediary for
>/w/ > /k/ -- I knew I had forgotten something. I haven't been able to
>find out what the accepted view is on the order of the changes,

I don't know that either. But it seemed to me that /d/ is somewhat closer
to /4/, while /w/ should have changed to /g_w/ (with a later devoicing,
like in the other non-aspirated voiced stops) rather than directly to
/k_w/, whence the proposed sequence of changes. But /tk(_w)/ > /4k(_w)/
isn't totally impossible, either.

>but I
>still think it looks better if the first stop is changed to something
>different before another is introduced:
>    */dwo/ > /L*wo/ > /L*g_wo/ > /L*k_wo/ > /rk_wo/ > /rku/ > /Erku/

I see. Yes, this seems probable, too. (OTOH two stops would be more prone
to dissimilation, IMO)

>where /L*/ is some alveolar liquid phoneme that can not be determined
>more precisely --- /d/ > /l/ as a direct step may be more common than
>/d/ > /r/.

It depends. For example, on the exact quality of /d/: if it was retroflex,
direct change /d/ > /r/ may be more natural (cf. American English).

BTW, it seems that Armenian does not confuse PIE /r/ and /l/. Or do I
forget something? At any rate, the examples of /r/ in place of IE /l/
which I've seen all looked like Iranisms (very common in Armenian).

>Lars Mathiesen (U of Copenhagen CS Dep) <[log in to unmask]> (Humour NOT