Re: Prevli phonology
|From:||Dirk Elzinga <dirk.elzinga@...>|
|Date:||Wednesday, January 16, 2008, 1:47|
I hesitated posting these comments because I'm leaving in the morning for a
conference in NYC, and I may or may not be able to read email while I'm
gone. But, no offense to those who are really into it, I just can't persuade
myself to be interested in prehistoric star charts and the mathematics of
I've been reading (and re-reading) your short description of Prevli
phonology. It is very interesting, and to this Americanist, very strange. I
recognize (I think) some things from Austronesian (no surprise there!), and
I find them refreshing. I had a few comments; take them or leave them, as
they are useful to you.
The whole document reads like a hybrid between a description and a plan of
action, which is what I suppose it is. But it makes for a rather uneven
tone, and some things I'd like to know more about are only briefly sketched.
Of course, it is clearly marked "preliminary version", so I'm sure this will
be worked out.
Since voicing and spirantization seem to occur in the same environments, I
would recommend using the cover term 'lenition' for both. You can then refer
to 'voicing' or 'spirantization' separately as needed.
How the lenition processes interact with metathesis is not clear to me.
Right now you have the sandhi grids which show what happens, but no real
explanation or even description of the processes involved. They look like
lenition, but not quite. One alternation caught my eye: /pb/ -> [v:]. This
seems highly unlikely to me; It seems more natural to have [p:] or [b:]. Is
there a natlang precedent for this?
How is the s-acute pronounced? I'm assuming that a comma indicates
palatalization, but I'm not sure about that, either, since in some cases you
have a superscript <j>, which appears to resolve hiatus in some vowel
clusters, and elsewhere you state that you're using IPA, except that <y> is
used for [j].
The vowel harmony processes for unstressed vowels in closed syllables sets
up a larger inventory of vowels in a "weak" position than you have in a
"strong" position; this seems unusual. I suppose that it can be rationalized
as "coloring" on an otherwise indeterminate vowel, but in that case, it
seems to me that you would to describe it as vowel reduction in stressless
syllables with low-level, surfacy phonetic changes rather than as a set of
The vowel harmony alternations for stressless vowels in final open syllables
seems much simpler, but I wondered about two alternations. You have /o-i/ ->
[o-e], but /e-i/ -> [e-i]; you have /e-u/ -> [e-o], but /o-u/ -> [o-u]. This
seems backwards to me, and I would have expected /e-i/ -> [e-e] and /o-u/ ->
[o-o], as well as (or instead of) /o-i/ -> [o-e] and /e-u/ -> [e-o]. Is
there a natlang precedent for your arrangement? I know that in Shona (?)
underlying high suffix vowels agree in height with a preceding mid vowel.
Finally, I'd be interested in the arguments for setting up the
unmetathesized (CVCVC) forms as underlying. I mean, *you* know that they are
because you're the one making up the language. But I don't have enough data
to come to that conclusion myself.
Thanks for the fun, and don't take any of these comments too seriously.
On Jan 12, 2008 3:42 PM, ROGER MILLS <rfmilly@...> wrote:
> For your delectation, I've uploaded a prelim. version of this at
> I'm not entirely happy with it, and feel I may have left out or glossed
> something important... any comments???