mode intelligent
open 'proto-vengic.txt' as LEXICON
C = p t c k q b d j g v z y ɣ f s x h m n ɳ ŋ l r w pʲ bʲ mʲ ʈ ɖ ʂ
A = a ã aː ãː
E = e ẽ eː ẽː
O = o õ oː õː
[IH] = ɪ ɪ̃ ɪː ɪ̃ː
[UH] = ʊ ʊ̃ ʊː ʊ̃ː
I = i ĩ iː ĩː
U = u ũ uː ũː
[VSHORT] = a ã e ẽ o õ ɪ ɪ̃ ʊ ʊ̃ i ĩ u ũ
[VLONG] = aː ãː eː ẽː oː õː ɪː ɪ̃ː ʊː ʊ̃ː iː ĩː uː ũː
[VORAL] = a aː e eː o oː ɪ ɪː ʊ ʊː i iː u uː
[VNASAL] = ã ãː ẽ ẽː õ õː ɪ̃ ɪ̃ː ʊ̃ ʊ̃ː ĩ ĩː ũ ũː
V = [VSHORT] [VLONG] ɐ ь ъ % reduced vowels
[PB] = p t c k q b d j g
P = p t c k q
B = b d j g
[SZ] = v z y ɣ f s ç x h
S = f s ç x h
% normalize ASCII
aa > aː
ee > eː
oo > oː
ii > iː
uu > uː
[VORAL]m > [VNASAL] / _C or _#
ng > ŋ
gh > ɣ
[VORAL]ɴ > [VNASAL] / _
% the sound changes in
http://www.incatena.org/viewtopic.php?p=1049472#p1049472 are pretty out of date
% it's assuming sesquisyllabic (CV-)CVC, but Proto-Vengic was actually strict (C)V
% but we can't take this structure as representative of Ziwanic, because Vynyi "is" Ziwanic, by some bizarre standard
% so... let's take the *N split as representative of Ziwanic
% (and fricative reflexes of V < *B + plain-voiced reflexes of B < *MB, but that's everything except Zotic)
% ( + possibly some other early branches)
ɳ > mʲ / U_
ɳ > ɖ
% Length shift to handle CVVCVCV, which is otherwise difficult
[VLONG]C[VSHORT] > [VSHORT]C[VLONG] / _CV#
% Length rule
[VLONG] > [VSHORT] / [VLONG]C_
% now that we've defined Ziwanic, let's get Proto-Everything-But-Vynyi
% which *is* defined by the sesquisyllabic structure
% from earlier penultimate stress
% however, unlike in Hluic, long vowels attract stress
% specifically, the rightmost long vowel
% as usual, we'll use ʜ to represent stress...
% first, stress every long vowel
% if you do 0 > ʜ / [VLONG]_, it won't work word-finally for some reason
[VLONG] > [VLONG]ʜ / _
% then remove all stresses but the rightmost
ʜ > 0 / (CV)+_(CV)+ʜ
% assign regular penultimate stress
0 > ʜ / [VSHORT]_C[VSHORT]# not ʜ(CV)?_C[VSHORT]#
% stress all monosyllables (that aren't already stressed)
V > Vʜ / #C_#
% Proto-Vengic root structure:
% (C1{ă ĭ1 ŭ1})(C2){a e o i u}(C3({ă ĭ2 ŭ2}))
% C1 = p t c k ʔ f s x h m n
% C2 = any C or Pr Fr or labial + y
% C3 = any C except unvoiced non-glottal plosives
% Nasality contrast is lost in unstressed syllables.
% How can we get those clusters? Well, let's say unstressed a > schwa, shorter than all the other vowels, and is lost.
% I'm not too sure about this tho - you get height/length interactions elsewhere in Vengic:
% - in Vynyi, clusters [probably] come from lost *high* vowels
% - in Whatic, final -a is less prone to loss than other final vowels
% But I'd originally planned the C1 vowels to condition height mutation, and who needs that.
% So prestress vowels won't be e o i u, but rather a i u.
i u ĩ ũ > 0 / {[PB] [SZ]}_rVʜ
i ĩ > 0 / {p b f v m}_yVʜ
% Long vowels will interfere with this somehow, of course...
% they'll give you... disesquisyllabic words?
% So let's assign secondary stress.
V > Vʜ / _C{C}?VC{C}?Vʜ
a ã > ɐ / ʜC{C}?_
e ẽ o õ > ɐ / ʜC{C}?_
i ĩ u ũ > ь ь ъ ъ / ʜC{C}?_
a ã > ɐ / _C{C}?Vʜ
e ẽ > ь / _C{C}?Vʜ
o õ > ъ / _C{C}?Vʜ
i ĩ > ь / _C{C}?Vʜ
u ũ > ъ / _C{C}?Vʜ
% One more consonant shift...
g > k / #_
g > ŋ
% Good enough.
ʜ > 0
pʲ bʲ mʲ > py by my
% Not sure what this is. Is it contrastive?
w > v
% More orthographic stuff
z > ð
% Oh, drop length while we're at it.
[VLONG] > [VSHORT]
close LEXICON as 'proto-ziwanic.txt'