Theiling Online    Sitemap    Conlang Mailing List HQ   

Phonology drift

From:John Vertical <johnvertical@...>
Date:Saturday, November 25, 2006, 21:51
Giv'n the recent talk on conlang diachronics, I think I'll actually try &
post this here insted of the ZBB. This is yet another sketchlang that may or
may not become anything, but is fun to play with.

So I have this here stock protolang which is the distant (maybe 5000 years
ago) ancestor of uwjge & a handful of sketches:

p' t' k' q' (=ejectivs)
p t k q
b d g
m n N
l r
s x
a i u

CCVVCC with obst+/l r/ initial clusters (two series wrt/ voicing in stops),
no restrictions on vowel combos, and sonor+obst final clusters (one series).

The sketch in question starts with the following changes:

p p' > f b'
t' > z
k' k g N > ?j c J\ J
q' q > k' k
li > L
r > G

The POA change of the dorsal stops might actually not be a shift as much as
an alternate output of a yet older vowel system collapsing but leaving
adjacent consonants affected; but N g would be expected to remain in that
case. However N > J (nigh-)universally might not be too bad, and g could
also > G... It might also be worth of note that lenition of ejectivs to
vricativs goes on in a few other branches as well. But at any rate, this is
just the "setup" part.

In the more questionable phase 2:

d J\ > r t'
s n l > T n_d l_d
c J L > s n l
b' > v

The main issue is the very asymmetrical fate of the palatal stops. I'm not
sure which of these two might be the better explanation:

J\ > c' > tS' > S' > s' > t'
c > tS > S > s

This invokes the rule that with voiced stops, back ones are more marked,
while the scale run the other way 'round with ejectivs; but while uvular or
velar voiced stops shifting to something else (or failing to emerge in the
first place) is reasonably common, I'm aware of neither a palatal nor a
spontaneous ejectivization precendent. A bonus, however, is that I could
lenite p' to v directly in the first phase with this roote; no need for the
implosiv intermediate.

J\ > d` > d' > t'
c > t` > ts` > s` > s

This has to do with d' usually being more backed than its pulmonic
counterpart; it has in many African languages shifted to a lone d`, but I
wonder if working the change backwards is plausible. The b' > v shift would
also have to be crammed in quite a small timescale then, but that's no dire
problem. Nor is the spontaneous retroflex affrication, which also happens in
at least one related language.

An issue tied with this choice is whether J L > n l would be more plausible
with or without a postalveolar intermediate. Or would it make any sense for
them to "jump" over into the *dentals* insted?

The final phase is labial loss: m b v f > N ? D h (or maybe N w D T)
which is probably areal influence or something. I alreddy blame the
influence of a neighboring labialless language for some changes in this
lang's closest (so far) relativ, so I figure I could pull the same trick

Anyway, obviously this is just an outline - the development of the consonant
inventory rather than of all individual phonemes. Frex r > G could go thru a
very English-like intermediate of r\` or r\, which could also trigger vowel
changes and retroflexion (if I choose to use'em); itself changing in the
process to stuff ranging from, say, s to N to zero. The point is, there is
not going to be any (or only very little - I'll probably work dental stops
in at least) extra phonemes emerging via the conditioned changes; the
large-scale evolution is completely based on these few universal or
near-universal shifts. So does this seem plausible? As you can see, a few
consonants drift way off from where they started... 5000 years is a long
time, but large-scale consonant shifts don't happen very commonly, do they?

John Vertical

Windows-työpöytähaun avulla löydät tiedot koneeltasi silmänräpäyksessä.


David J. Peterson <dedalvs@...>