Proto-Ranic


I think this is about as far as I'm going to take this. I've spent about two weeks on it, and though I'm sure I could go farther, it's at a good breaking point and it's getting in the way of other projects.

I'm afraid I skipped over all the grammars, concentrating only on the lexicon.

For reference, the full names of the languages and their creators:

The first stage was to collect cognates. You folks didn't make this very difficult. :) I was a good deal meaner with Fáralo: I probably changed the meanings of a third to a half of the vocabulary. Even that probably doesn't reflect the level of meaning change in a natlang.

I respelled the words as I was copying them, partly to downplay dissimilarities due to different spelling conventions, but mostly so I could use my Maraille font throughout. Maraille doesn't have àèìòù, which is why these become āēīōū in Mao. It also doesn't have æ, so I wrote ae instead. This may have been a mistake, since it makes it look too much like ao.

I also put aside the net of borrowings that you all worked so hard on. :) A few came in anyway and caused some trouble.

The next stage was to analyze initial consonants, then medials, then vowels, looking for correspondence sets.

Next, I looked through the correspondence sets, seeing if they could be related, and assigning proto-phonemes.

Then, I went through the initials/medials/vowels again, comparing the cognate sets to the reconstructed forms; this is where things started to feel like they were coming together.

Below you can see my spreadsheets, with the reconstructed forms. With each correspondence set I put the reconstruction and sound changes, and I've added some notes on my reasoning.

Initials

Initial vowel

ENGLISH THOK MUTS MAO ZELE PROTO CHANGES CHINESE
ear itta ?ita īzā ida
lead (vb) îtte iinda īndä
night utta ûta uda
1s a ey
smell îhe īošy imši
and u y` ü
blow une unu unu
breath, wind uhiphi ūšipí roazh ? ušipi huxi
carry upe āopy` uwp?
chest mpu āopú
never uhôhap ūúšae uhušay
smooth ûhamyu isāmīo üsamyu
strong unî āonyš uwni
thin ume ūma umä
throat āowyš wiš
... ... 0

Most of these probably actually start with *s or *š, but since we don't have Muts or Zele cognates I can't tell. It doesn't look like the protolanguage had many initial-vowel words, possibly none.

Velars *k, *g

some gimsuh giusūš neafuri gimsuh Thok g → k
bone ku gu gu Thok g → kh / _i gu
root kunam guuna gunao gūnam genben
lie down, rest komô gwuh gaomūš gomuš Zele g → y / _i
love khîṫe ktsan gitšaē yizã gitzan Zele g → h otherwise?
fly kome gaomā home gomä
k g g h, y, n
bite kajoh kalāoš káyaw Thok k → kh / _i
mountain ke kae key
hill kejdi kaezí keydi
steal kejpuns kaépyš keypuš
new kahake khaxa? kašága kašageh
hand, year ksah kazāš kazaš
horn ku ku
winter ku?û kulú kuyu
crops kôhi kuhsi kus kussi
spirit kwi kuu ky` kwe gui
herd kuni kuunii kylí kūnyī
nut, seed kuxî kxam kumC
brother kakam kxan kakam
month, moon kyumi kyywi kiomi kyumi
mouth ka kāoš kaw kou
fall kaheṫe kašaētīo kašeyt-
suck kapâ kamá kabá
four kata kazā kada
five katâhi azāš kada-ši
six katâhihwi zāšys kada-šišwe
seven katâṫe kataé kada-tey
eight katâṫyu (īka) kada-dyu
nine katâṫîhi āgazāš kada-di-ši
ten katâṫîhihwi gašys kada-di-šišwe
eleven katâṫîṫe zātaé kada-di-tey
cold kuṫyô kutšūlū kuzeyu kutzuh
temple kuha kūpas kupa
priest kuhu kuhu
woman khîphî kiispy kispio kīspü
light khikîtû kigiozi kigüdü
k k k k
follow xêṫe kanta kant- Thok k → x / _FCC gensui ?
man xêma kemma kaomao 'husband' kemmaw
heavy xemîtû kaemiozī keymītü
x k k

As you can see, reconstructing *k is straightforward. *g is almost as easy-- it's preserved in two languages, and Zele's g → y suggests a voiced sound. So there wasn't much warrant to worry about other possible constrasts, such as aspiration. (Thok does have an aspiration constrast, but it's pretty clearly conditioned by the following vowel.)

I'm not sure what's going on with Thok's x— it seems to occur only in checked syllables with front vowels, but there's not really enough instances to be sure.

Say, how did Chinese get there? I noticed a few glaring borrowings, notably mao and kou. So I looked up every word to see if there were more. I'm sure I missed more.

Nasals *m, *n

wheat mâta mata máza matazh máda no changes! -
fur ma me mao maw mao
hear, listen mîṫe mets matsīo mesea metsew
flax maṫi metsi matsi metsi
small mimatû mi mi
tail mipa mifa mipā mipa
rain mim mii mio mīm
family miintsu mytsú measuhi mīntsu
sheep mima miiwa mima mimazhi mīma
water mîkhi mimki miogí meakihi mimki
eye mu mu mu -
clothes mwipa mûfa mupā mepazh mwepa
argue mwitâ mûsa? mutá mwetah
think mwîṫû mûtsi? mutsí mwetsü
mushroom muhu muuhu mūšu
mother mama mwa ma mama
grass minam mwina mināo minam minam
sit mome mwun maomy` momun
face myumyu mywyh miomīoš myumyus mian
m m m m
person nam na nao nam nam
autumn nahupu nesufu nasumú nesubu
do ne 'be' ni ni
spring nihipa niisifa nizipá nīzipa
green nime nin nināo ninen
want nittû nit nizīo nidu-
summer nofa naomaē nobäy
sew nwihe nûs nusīo nwese
cow nona nynoh non?
give nysa? niotá nyuta
turn nattô nazāo nadaw
noodle nayu nalū nayu
tie nuwī nuwi
pull nyô nizīo nidum ?
n n n n

No mysteries with *m or *n.

Approximants *y, *w

drink i lio yi Thok y → 0
pine jasuwy lásuwy yasuwi Muts y → j, û before u
black uttû juuti? ludī yūdih Mao y → l
rice nyaki njax lāgī yagi
kick û yu
deer uka ûxa lukā 'reindeer' yuka lu
feather uma lumāo yumaw yumao
fear ohe joh yoše
0 nj, j, 0 l
if ?wu? ūmú wumuh Thok w → 0
earth, dirt uhî uuho?o wušāowá wūšowo
flow uunsa wysāo wūnsaw Muts w → u / _u
star uni wejnih waniš benir weniš
bad an wen wanu van wen Zele w → b (/ v)
wolf wîhî wiatsu wíatsū boabun wihatsü
freeze wûwi? wumí wumih
dig ahô ys wīs wüs
garden îčî wyty wintü (yin)dao
dog îhî wyš wuš
0, w w, 0 w b, v

The first set is a puzzler. Muts is a big mess; no help from Zele— I reconstructed *l at first, and this is certainly defensible. Mandarin yumao was too big a clue, though, and it helped point out that the normal Muts reflex is also y. (As well, if there's a *w, it seems likely that *y would exist too.)

'Rice' doesn't fit in much and I suspect it's a borrowing.

*w was easier, since it's retained in two languages, and in three medially.

'Wolf' is just a mess wherever it shows up.

Labials *p

hair pa pa Thok p → f / _ï
beautiful ipa pa?a palā paya Thok p → ph / _ī,u, ü
sand pah pāš par paš Ind. pasir
few pimô piiwu pimu pīmuh
house nêpaha psa pasaē pasay
say ôtâ (?) psa? pazá pota podah
long psoh pusāoš pusaws
breast, chest pômî puhmun pušmyš pušmin
fish pumu pwu pumy` pumu
many pomî pwun paomy pomü
but pahu py?o?
climb pâhâ pasaēš pasay pa
call pyutâ piotá pyutá
brother púaswī
arm fêpam piipaw pípao pïmpam
back fêpu pípū pibuzh pïmpu
nose phîṫi piitsi pitší pītzi
sow phîṫâ ptsun putsyš putsun bozhong
skin phînî puunnis pynis pūnnis
p p p p

Not much to see here. I reconstructed *b medially, but can't find any evidence for it initially.

Sibilants *s, *š

shine čîmâ hap āma šama Thok š → 0
laugh hihe haxih āšišā šašïše (xiao)
happy ku hex āku šeku Muts š → h (yukuai, xingfu)
flower onam hjona ālaonāo šayonam Jp. hana?
horse uha hyha āoša šuwša Mao š → 0 + length on a
cliff hemna aoná šemna
live ixe hexa aēkā reke šeykä
fox wiṫî hu?ytsim šwetsim (huli)
kill wihâ hûsam ysaoš šwesa shasi
young uhu huuhu šūhu
die wime hwun ymy remoa šwemo xiaomo
often ihehyô aešae rer šeysu
red (pe) hafa? āmā rapa šabeh
sun, day, east (pa) hefa āma rabazhi šeba ri
west (phî) hefa?ii rababuhi šebahī
fire hefatsa amatsá rabasahi šebatse
hot (paṫû) hefatsi? āmatsi rabasi šebatsih re
warm (kape) hexafa? āgamā ragapa šegabeh
heart im īo šim xin
two ihwi īšys šišwe
0 h : r
good ?i ši Not enough to base s.t. on
eat si? ši chi
beetle sosiwi šūšípi foshibi 'locust Can't be cognates, medials
don't fit. Borrowing
right se á se Thok s → 0
snake xata seh āš seš
cloud pu sepu āpú sepu Mao s → :
other sesa āza shazhi sazä
all sesowo āzāwáo sezowo
south khim sexi āgio segim Zele s → f / _u
left (dir) sii í otherwise → sh
lion îthi siitsi sītsi
north ikhim siixi īgio shigimi sīgim
see îtû simti? iotí sheati simtih
bear child sintsih ytšiš sintzi
wash ûhe sis ís süs-
wrist wam so?on āowaē sowan
louse sufimsin ūpīos supims-
sleep, dream upû sufuf ūmū fufub subub
hate uhâ suha? ūšá sušah
mouse uhu sus 'rat' fush sus
tongue ôta suut útao sūtaw
bug yuhi sysi iozīo syuzü
play sysi? īosí syusih
fog, mist sysimma iomā syusimma
well, mine ṫwihi tssi saēšis 'mine' seyši
0 s 0 sh

The first set is weird: h and r? Medially the correspondence is h/h/š/r, which puts *š in the ring. I had *r for awhile, but *š fits xiaomo, xin and also coheres with the next set, *s. It's more attractive to say that initial *s and *š are lost in Thok/Mao than the unrelated *r and *s.

*s is easier, especially given the medial set h/s/s/š. Medially there's evidence for *z, but if there was initial *z it's merged with *s.

By the way, north/south are obviously derivatives of left/right.

'Beetle' is tantalizingly close to reconstructible (perhaps *sosipi), but Mao doesn't fit. So somebody's borrowing.

Similarly, I can't make anything of 'good' and 'eat'; not enough data.

Dentals *t, *d

far to ta taw Zele t → ts
vomit tâṫô tatsuh tatšūš tatzuš
fight taxeh takaēš takeyš Thok t → t / _e, ä
head ta te tao taw (except 'head') tou
so that te? teh Thok ts → č / _un
lentil tophî tfinh topinš
daughter tii ti
sky twam tiihen tílae tīyen tian
drop tja taelāo tayaw
pea to?uu taowú towu
go tohe tos taozāo tsoshã tozam
slow tome towa? taoma tomeh
come tsun tiosy` tyusun
knee twîtham tts tytsāo twetsam
child ṫêṫi ttsi tatsi tetsi
leg tu tu ty` tuy tui
seed tyam ?e tylāo tweyam
belly twî ?iih tylīš tweyī duzi
stop twixe tûk tygīo tseyea tswegim
enemy tuts tutā tut-
spit tôṫô tutsuh tútšūš tutzuš tutan
take tume twin tumy` tumi
near tyu ty tyü
fast tyume tywa? tioma tyumeh
2s tu tu
girl tyukam tiokāo tyukam
three ṫe te taé tey
twelve tôkata taéga tey-kada
compare ṫehâ tesa? taesá teysah
sing ṫêṫamâ ttsawa? tatšāmá sumazhi totzámah
egg ṫênwam taenūwaē teynuwan dan ?
son te ta te
white čîmê tunmam tymao tunmäm
snow čîmî tuunmam tymáo tūnmam
t t t ts
who suu zu sū ? Thok d → tt (exc. 'know' ṫ)
swim ttahapû tsafi? zapí dasafih
know ṫêṫâ ttsa? zaotšā toza dowtzah Muts d → t before lost vowel zhidao ?
silver fêṫuma zumao dumaw
hit ttatô zatūš tatu datuh da
Mao d → z
father ttatta data zašā dada Zele d → t
sea duu tuzh
river ttwiku dûûxu zugú dwegu
tt d, ts, s z t

*t is pretty straightforward, though Thok's č in two words is strange. It may be related to the k → x change, and also seemed to occur only in checked syllables.

I'm not very confident in that *d. The medials are no help: the distinction from *t is lost except for Mao, which has z. I reconstructed *z at first; but *d seems to fit the phonological facts better— the reflexes are almost all dental stops, and *d → t seems a little more plausible than *z → t. And when it seems clear that there was a *g, then I'd rather expect *b, *d as well.

Affricates *ts, *tz

neck ṫîpu ( hy?uun ) tsikamá Thok ts → ṫ
copper tsnohi tsunūi tsunohi
hide tsamm tsaomī tsammi
wing ṫêpam tsišpāo tsišpam chibang
squeeze ṫahe tsah tsašāo tsašew Zele ts → t zhaqu
ash (paṫâ)ṫačî tsaty tsazy tsadi
smoke ṫattyu tsazío tsadī
elbow ṫêṫu tsotsu tsaotsú tsowtsu zhou
hunt ṫokû tsox tsaokīo tsokü zhuigan
tree ṫonam tsna tsaonāo tsonam
branch ṫopi tsfi tsaopí tsopi
sister ṫêṫam tsan tsatsae toza tsotsay jiejie
guts ṫêmi tsemmi tsaomi tsemmi
forest tse tsao tsaw
then tse tsaē tsey
fat tjems tsipaē tseyp-
finger tsii tsi tsī (shou)zhi
sharp ṫe tsy tsï
frog tsimho tsiošwā tsimšV
foot ṫu tsu tsú tsu jiao
pig tsuutsu tsú tsū zhu
stone ṫuma tsuuwa tsuma tsūma
pick ṫipô tsfuh tsipuš
ugly ṫînu tsuunv tsynu tsūšu
blood ṫûṫi tsiitsi tsutsï
farm tsuntin tsyzyš tsuntu
beech ṫwî tsy?uun tswehun ?
ts ts t
cut tsuhkin tšušky` tzuškin See medials: def. phonemic
or ṫahu tsa tša tza-
throw ṫohî tssams tšaosāos tzosams
bird ṫîphî tsiifi tšímī tzībī
metal ṫumaxa tšimaetú tzümayt-
split ṫômô tsuhma tšušmāo tzušmaw
ts

*ts seems pretty good, when two languages retain it and the others are at least compatible with it. It could have been some other affricate, of course, like *č.

The medials make it clear that some sort of constrast with *ts is going on: Mao has ts vs. tš and Zele has s vs. z. Consider *tz to be just a recognition of the contrast without much attempt to define it. It could equally well be *tš or *dz or *dž.

Medials

Velars *k, *g

ENGLISH THOK MUTS MAO ZELE PROTO CHINESE CHANGES
new kahake khaxa? kašága kašageh Thok g → k
light khikîtû kigiozi kigüdü Thok g → kh / _i (exc. 'rice')
rice nyaki njax lāgī lagi Thok g → x / _e ??
warm kape hexafa? āgamā ragapa šegabeh
south khim sexi āgio segim Muts g → x
north ikhim siixi īgio shigimi sīgim
stop twixe tûk tygīo tseyea tswegim Zele g → y / _i (exc. 'north')
river ttwiku dûûxu zugú zwegu
k x g g, y
live ixe hexa aēkā reke šeykä Thok k → x /_ä
fight taxeh takaēš takeyš
hunt ṫokû tsox tsaokīo tsokü zhuigan Muts k → x
girl tyukam tiokāo tyukam
deer uka ûxa lukā 'reindeer' yuka lu
brother kakam kxan kakam
happy ku hex āku šeku
twelve tôkata taéga tey-kada
k x k k

Again, it seems pretty easy to reconstruct *k and *g. As we'll see, Muts likes to fricativize medial stops.

We saw Zele g → y with the initials as well, though I'm quite hazy on the exact condition.

Approximants *h, *y, *w

priest kuhu kuhu Muts h → ? after y
but pahu py?o? p?hu
young uhu huuhu šūhu Mao h → 0
eat si? ši ših chi
copper tsnohi tsunūi tsunohi
so that te? teh
wolf wîhî wiatsu wíatsū boabun wihatsü
h ? 0
bite kajoh kalāoš káyaw () yao? Thok y → 0 exc. seed, noodle
winter ku?û kulú kuyu Muts y → ?
noodle nayu nalū nayu but sometimes → j
beautiful ipa pa?a palā paya Mao y → l
drop tja taelāo tayaw
sky twam tiihen tílae tīyen tian (similar to initials)
seed tyam ?e tylāo tweyam
belly twî ?iih tylīš tweyī duzi
flower onam hjona ālaonāo šayonam
beech ṫwî tsy?uun tsweyun ?
0 ?, y l
all sesowo āzāwáo sezowo Thok w → 0 adjacent to u
pine jasuwy lásuwy yasuwi
tie nuwī nuwi
egg ṫênwam taenūwaē teynuwan dan ?
throat āowyš wiš
w w w
earth, dirt uhî uuho?o wušāowá wūšowo Muts owV → o?V
wrist wam so?on āowaē sowan wanzi
pea to?uu taowú towu
w ? w
argue mwitâ mûsa? mutá mwetah Thok h# → ^
say ôtâ (?) psa? pazá pota podah exc. eh# → e
hate uhâ suha? ūšá sušah
compare ṫehâ tesa? taesá teysah Muts h# → ?
know ṫêṫâ ttsa? zaotšā toza dowtzah zhidao ?
sing ṫêṫamâ ttsawa? tatšāmá sumazhi totzámah Mao h# → |
red pe hafa? āmā rapa šabeh exc. ej# → ā or a
warm kape hexafa? āgamā ragapa šegabeh
new kahake khaxa? kašága kašageh Zele h# → 0
slow tome towa? taoma tomeh
fast tyume tywa? tioma tyumeh
hot paṫû hefatsi? āmatsi rabasi šebatsih Could also be -y or -w!!
black uttû juuti? ludī yūdih
think mwîṫû mûtsi? mutsí mwetsih
see îtû simti? iotí sheati simtih
play sysi? īosí syusih
swim ttahapû tsafi? zapí dasafih
freeze wûwi? wumí wumih
if ?wu? ūmú wumuh
^ ?# | 0

That first set is pretty unsatisfying. Initial *h in Thok may also reflect *s/*š, but the reflexes would be different in Muts/Mao. *h is just a pis aller since there seems to be something there, but we can't say what. (Both h and ? are at the bottom of the lenition hierarchy, so the original could be almost anything.)

*y was discussed above. At least *y makes 'sky' *tīyen, closer to tian than *tīlen would be.

*w is fairly straightforward. If we had more data, we might get a more likely-sounding condition on Muts w → ?.

The last set could be various things. As we'll see, Thok's circumflexed vowels often indicate something elided, but doesn't say what it was. Muts's ? could be anything. Mao usually has a compensatory high tone, though again this doesn't tell us what was lost. I picked *h because it more or less matches the first set, but it could also be *w or *y or God knows what. But I did reconstruct diphthongs elsewhere, and they normally didn't generate a ? in Muts.

Nasals *m

sheep mima miiwa mima mimazhi mīma Muts m → w
many pomî pwun paomy pomü
few pimô piiwu pimu pīmuh
fish pumu pwu pumy` pumu
die wime hwun ymy remoa šwemo xiaomo
sing ṫêṫamâ ttsawa? tatšāmá totzámah
stone ṫuma tsuuwa tsuma sumazhi tsūma
take tume twin tumy` tumi
freeze wûwi? wumí wumih
if ?wu? ūmú wumuh
metal ṫumaxa tšimaetú tzümayt-
thin ume ūma umä
smooth ûhamyu isāmīo üsamyu
feather uma lumāo yumaw yumao
silver fêṫuma zumao dumaw
fly kome gaomā home gomä
lie down, rest komô gwuh gaomūš gomuš
sit mome mwun maomy` momun
month, moon kyumi kyywi kiomi kyumi
slow tome towa? taoma tomeh
fast tyume tywa? tioma tyumeh
face myumyu mywyh miomīoš myumyus mian
m w m m
root kunam guuna gunao gūnam genben Muts -m → 0 (exc. 'arm')
rain mim mii mio mīm
grass minam mwina mināo minam minam Mao -m → -o
person nam na nao (agentive) nam nam
arm fêpam piipaw pípao pïmpam
flower onam hjona ālaonāo šayonam
north ikhim siixi īgio shigimi sīgim egg' is odd
egg ṫênwam taenūwaē teynuwan dan ?
tree ṫonam tsna tsaonāo tsonam
girl tyukam tiokāo tyukam
heavy xemîtû kaemiozī keymītü
brother kakam kxan kakam
wing ṫêpam tsišpāo tsišpam chibang
m# 0# o# m#
nut, seed kuxî kxam kumC
water mîkhi mimki miogí meakihi mimki Thok mC → ^C
man xêma kemma kaomao 'husband' kemmaw
face myumyu miomīoš myumyu Mao mC → oC
fog, mist sysimma iomā syusimma
hide tsamm tsaomī tsammi
guts ṫêmi tsemmi tsaomi tsemmi
cliff hemna aoná šemna Zele imC → eaC
throw ṫohî tssams tšaosāos tzosams (C → f in neafuri)
louse sufimsin ūpīos supims-
some gimsuh giusūš neafuri gimsuh
frog tsimho tsiošwā tsimšV
smell îhe īošy imši
see îtû simti? iotí sheati simtih
^C mC oC aC
shine čîmâ hap āma

Medial *m is a little more interesting, since it's weakened in Muts.

Final *m is more of a puzzle. Mao has a bunch of ao, which I've ended up tracing to three separate sources: *aw, *o, *m. Why not all the same? Largely because we actually see the final m in Thok, Zele, and (before a consonant) Muts; where we don't, I reconstructed *aw or *o.

The protolanguage doesn't seem to have tolerated many medial clusters, but there was plenty of evidence for *mC, *nC, *sC. It soon became clear that Muts had kept the medial nasal clusters, Thok replaced them with a circumflex on the previous vowel, Mao turned them into o, and Zele into a. Maybe I was primed to see consonant → o changes since I'd used them in Fáralo!

I left 'shine' in just to show that I didn't know what to do with it. None of the consonants really fit an existing pattern; very likely this isn't a cognate set at all.

Nasals *n

root kunam guuna gunao gūnam genben no change
grass minam mwina mināo minam
cow nona nynoh non? muniu
flower onam hjona ālaonāo šayonam
egg ṫênwam taenūwaē teynuwan dan ?
tree ṫonam tsna tsaonāo tsawnam
copper tsnohi tsunūi tsunohi
ugly ṫînu tsuunv tsynu tsūšu
blow une unu unu
strong unî āonyš uwni
bad an wen wanu van wen
star uni wejnih waniš benir weniš
herd kuni kuunii kylí kūnyī
n n n n
love khîṫe ktsan gitšaē yizã gitzan Thok n# → ^ (but → 0 for e)
come tsun tiosy` tyusun Mao n# → -V, but lost after u
green nime nin nināo ninen Zele n# → ~
sow phîṫâ ptsun putsyš putsun bozhong
wrist wam so?on āowaē sowan wanzi
sky twam tiihen tílae tīyen tian
cut tsuhkin tšušky` tzuškin
breast, chest pômî puhmun pušmyš pušmin
0 n V ~
lead (vb) îtte iinda īnde (yin)dao Thok nC → ^C (but e doesn't → ê)
follow xêṫe kanta kant- gensui ?
garden îčî wyty wintü
farm tsuntin tsyzyš tsuntu
snow čîmî tuunmam tymáo tūnmam Mao nC → C and vowel → y
white čîmê tunmam tymao tunmäm
skin phînî puunnis pynis pūnnis pi Zele inC → eaC
bear child sintsih ytšiš sintzi but → s here; cf. neafuri above
family miintsu mytsú measuhi mīntsu
lentil tophî tfinh topinš
flow uunsa wysāo wūnsaw
^C nC C aC

Much like *m, we have clear medial *n and special cases for final *n and *n before a consonant.

I'm not sure what's happening with Thok wam and twam— perhaps these changed randomly to -m. If we had more examples a pattern might emerge.

Labials *p, *b

steal kejpuns kaépyš keypuš Thok p → ph / _i
temple kuha kūpas kupa spirit house
cloud pu sepu āpú sepu Muts p → f
carry upe āopy` uwp? but not where previous Mao vowel
chest mpu āopú has a tone!
arm fêpam piipaw pípao pïmpam
back fêpu pípū pibuzh pïmpu
tail mipa mifa mipā mipa weiba
clothes mwipa mûfa mupā mepazh mwepa
spring nihipa niisifa nizipá nīzipa
louse sufimsin ūpīos supims-
lentil tophî tfinh topinš
branch ṫopi tsfi tsaopí tsopi
pick ṫipô tsfuh tsipuš
swim ttahapû tsafi? zapí dasafih
fat tjems tsipaē tseyp-
p f p p, b
beetle sosiwi šūšípi foshibi 'locust Thok b → p
suck kapâ kamá kabá Thos b → ph / _ī
autumn nahupu nesufu nasumú nesubu
summer nofa naomaē nobäy Muts b → f
red pe hafa? āmā rapa šabeh Mao b → m
sun, day, east pa hefa āma rabazhi šeba ri ?
west phî hefa?ii rababuhi šebahī
fire hefatsa amatsá rabasahi šebatse Zele s.t. → p instead
hot paṫû hefatsi? āmatsi rabasi šebatsih
warm kape hexafa? āgamā ragapa šegabeh re
bird ṫîphî tsiifi tšímī tzībī
sleep, dream upû sufuf ūmū fufub subub
p f m b, f

*p is easy enough; one again Muts weakens it medially.

The second set was a puzzle. Where'd that m come from in Mao? For awhile I considered *mp; but the pattern is unlike anything that happens with other *mC. *b seems warranted by Zele, and also fits in with having a *g, above. And *b → m for Mao seems plausible.

Nonetheless Zele doesn't really have p/b in the pattern I'd expect. Rapa for 'red' is hard to explain.

'Beetle', as noted before, just doesn't fit in; not all the words can be cognates.

Sibilants *s, *z

pine jasuwy lásuwy yasuwi Thok s → h
autumn nahupu nesufu nasumú nesubu s.t. with ^ before ...or after!
sew nwihe nûs nusīo nwese
climb pâhâ pasaēš pasayC pa Zele s → sh
house nêpaha psa pasaē pasay
long psoh pusāoš pusaws
kill wihâ hûsam ysaoš šwesa shasi
mouse uhu sus 'rat' fush sus
play sysi? īosí syusih
compare ṫehâ tesa? taesá teysah
wash ûhe sis ís süs- xi ?
smooth ûhamyu isāmīo üsamyu
dig ahô ys wīs wüs
come tsun tiosy` tyusun
throw ṫohî tssams tšaosāos tzosams
one uhô usü
^h s s sh
crops kôhi kuhsi kus kussi
woman khîphî kiispy kispio kīspü Thok sC → ^C
^p sp sp, šp
all sesowo āzāwáo sezowo Thok z → h
hand, year ksah kazāš kazaš Muts z → s
spring nihipa niisifa nizipá nīzipa Zels z → zh (exc. 'go')
other sesa āza shazhi sazä
go tohe tos taozāo tsoshã tozam
bug yuhi sysi iozīo syuzü
pull nyô nizīo nidum ?
swim ttahapû tsafi? zapí dasafih
h s z š, ž

*s is fairly straightforward, especially as Mao retains it instead of (as with the initials) losing it.

It seems like clusters *sC were allowed too, though for some reason there aren't many of them.

In the second set, both Mao and Zele suggest that the difference from *s is one of voicing. It could have been something else, but if no there's no evidence what it could be.

Sibilant *s

five katâhi azāš kada-ši
six katâhihwi zāšys kada-šišwe
ten katâṫîhihwi gašys kada-zi-ši
often ihehyô aešae rer šeysu
two ihwi īšys šišwe
fall kaheṫe kašaētīo kašeyt-
sand pah pāš par paš
laugh hihe haxih āšišā šašïše
horse uha hyha āoša šuwša
snake xata (?) seh āš seš she ? Thok š → h
hate uhâ suha? ūšá sušah
fight taxeh takaēš takeyš Muts š → h
never uhôhap ūúšae uhušay š → x / a_, ä_
breath, wind uhiphi ūšipí roazh ? ušipi
star uni wejnih waniš benir weniš Zele š → r
dog îhî wyš wuš
earth, dirt uhî uuho?o wušāowá wūšowo
new kahake khaxa? kašága kašageh
laugh hihe haxih āšišā šašïše
squeeze ṫahe tsah tsašāo tsašew
mushroom muhu muuhu mūšu mogu
fear ohe joh yoše
vomit tâṫô tatsuh tatšūš tatzuš
h h š r
well, mine ṫwihi tssi saēšis 'mine' seyši jingshui ?
h s š
cut tsuhkin tšušky` tzuškin
breast, chest pômî puhmun pušmyš pušmin
split ṫômô tsuhma tšušmāo tzušmaw
wing ṫêpam tsišpāo tsišpam chibang
^C hC šC

See the section on initials on why *s rather than *r. (Mandarin mogu is enough to make me wonder if it's something else entirely— maybe *h or *x— but certainly not *g!)

'Well, mine' is probably a red herring— none of the consonants, initials or medials, really fit.

Dentals *t, *d

give nysa? niotá nyuta
seven katâṫe kataé katey
call pyutâ piotá pyutá
tongue ôta suut útao sūtaw shetou
metal ṫumaxa tšimaetú tzümayt-
enemy tuts tutā tut-
eleven katâṫîṫe zātaé dátey
hit ttatô zatūš tatu datuh da
t t t t
five katâhi azāš (k)adáš Muts d → s when Mao has á after...
four kata kazā kada else d → t
hill kejdi kaezí keydi Mao d → z
light khikîtû kigiozi kigüdü Zele d → t
wheat mâta mata máza matazh máda
say ôtâ (?) psa? pazá pota podah Thok d → t exc. → ṫ before _i
eight katâṫyu īka kada-dyu
nine katâṫîhi āgazāš kada-di-ši
ash (paṫâ)ṫačî tsaty tsazy tsadi
t t z t
turn nattô nazāo nadaw
smoke ṫattyu tsazío tsadī
want nittû nit nizīo nidu-
night utta ûta uda
father ttatta data zašā dada Thok d → tt
black uttû juuti? ludī yūdih except sometimes it's t or č!
ear itta ?ita īzā ida
tt t z

I'm not very happy with *d; see the Initials section for the reasoning.

It looks like *d → t or tt randomly in Thok. On the other hand, four instances of the first set are really one word, *kada 'four', 'ash' is yet another reflex, *c, and 'say' is weird anyway because of its lost initial. That leaves just 3 clear instances of *d → t: kata, khikîtû, mâta. I still can't see a pattern, but it makes *d → tt predominate.

Affricates *ts, *tz

flax maṫi metsi matsi metsi Thok ts → ṫ
hear, listen mîṫe mets matsīo mesea metsew exc. 'lion'
think mwîṫû mûtsi? mutsí mwetsih
sow phîṫâ ptsun putsyš pütsün bozhong Zele ts → s (exc. toza)
hot paṫû hefatsi? āmatsi rabasi šebatsih (boabun is very weird)
fire hefatsa amatsá rabasahi šebatse
fox wiṫî hu?ytsim šwetsim y
lion îti siitsi sītsi shizi
child ṫêṫi ttsi tatsi tetsi zi
elbow ṫêṫu tsotsu tsaotsú tsowtsu zhou
sister ṫêṫam tsan tsatsae toza tsotsay jiejie
blood ṫûṫi tsiitsi tsutsï
knee twîṫam tts tytsāo twetsam
wolf wîhî wiatsu wíatsū boabun wihatsü
family miintsu mytsú measuhi müntsu
ts ts s
love khîṫe ktsan gitšaē yizã gitzan Thok tz → ṫ
cold kuṫyô kutšūlū kuzeyu kutzuh Muts tz → ts
nose phîṫi piitsi pitší pītzi Mao tz → tš
sing ṫêṫamâ ttsawa? tatšāmá totzáma Zele tz → z
vomit tâṫô tatsuh tatšūš tatzuš
spit tôṫô tutsuh tútšūš tutzuš tutan Could be dz or tš as well.
know ṫêṫâ ttsa? zaotšā toza dowtzah zhidao ?
bear child sintsih ytšiš sintzi
ts z

Not much to add that wasn't covered under the initials, except that Zele provides some evidence for a voicing contrast.

Vowels

The vowels were a major pain in the neck. In fact, the first time I looked at them I could hardly make anything of them— there were a huge number of correspondence sets, and no clear *e or *o cognates at all... it looked like the protolanguage had either three vowels or twelve. There was also so much data that it was very hard to get a handle on.

Looking at them again, I realized that I was paying too much attention to Mao. The other three fit in pretty well with a five-vowel system. And once that was sorted out, some of the other correspondence sets melted together.

The tables below are for the first vowel in the word only; restricting the analysis to these made the data much more tractable. I started a spreadsheet for the later vowels; but looking it over, it didn't seem that a full analysis would add much that I didn't already know.

Vowel *a

ENGLISH THOK MUTS MAO ZELE PROTO CHANGES CHINESE
father ttatta data zašā dada
swim ttahapû tsafi? zapí dasafih
hit ttatô zatūš tatu datuh da
suck kapâ kamá kabá
four kata kazā kada
five katâhi azāš kada-ši
six katâhihwi zāšys kada-šišwe
seven katâṫe kataé kada-tey
eight katâṫyu īka kada-dyu
nine katâṫîhi āgazāš kada-di-ši
ten katâṫîhihwi gašys kada-di-šišwe
eleven katâṫîṫe zātaé kada-di-tey
brother kakam kxan kakam
new kahake khaxa? kašága kašageh
fall kaheṫe kašaētīo kašeyt-
bite kajoh kalāoš káyaw
hand, year ksah kazāš kazaš
mother mama mwa ma mama
turn nattô nazāo nadaw
noodle nayu nalū nayu
house nêpaha psa pasaē pasay
climb pâhâ pasaēš pasay pa
sand pah pāš par paš Ind. pasir
beautiful ipa pa?a palā paya
other sesa āza shazhi sazä
red (pe) hafa? āmā rapa šabeh
shine čîmâ hap āma šama
laugh (hihe) haxih āšišā šašïše (xiao)
flower (onam) hjona ālaonāo šayonam Jp. hana?
fight taxeh takaēš takeyš
child ṫêṫi ttsi tatsi tetsi
vomit tâṫô tatsuh tatšūš tatzuš
drop tja taelāo tayaw
ash (paṫâ)ṫačî tsaty tsazy tsadi
smoke ṫattyu tsazío tsadī
squeeze ṫahe tsah tsašāo tsašew zhaqu
or ṫahu tsa tša tza-
rice nyaki njax lāgī yagi
wheat mâta mata máza matazh máda -
hair pa pa
pine jasuwy lásuwy yasuwi
a a a a
chest mpu āopú a + consonant
hide tsamm tsaomī tsammi Thok: an → ê
person nam na nao nam nam
follow xêṫe kanta kant- gensui ?
aC aC ao am

It' s not hard to reconstruct all these as *a.

We see most of the numbers here. The grammarians of Muts and Zele seem to be my least favorite type: those that leave out the numerals. And Mao's numbers are very worn down, to the point that it wasn't clear what consonants and vowels were left, compared with the Thok forms. Examining them separately, I saw that 4-12 were built on a simple pattern all from *kada 'four'. 5-7 are just conjoints of 4 and 1-3; 8-11 add another element -dyu/di in between; and 12 is just 'three-four', *tey-kada.

I can't say I've figured out where all of Mao's tones come from. I note in the sound changes column when I've found a pattern— for instance, when initial *s is lost, the next vowel gets a high tone.

By the way, scanning these sets, don't be misled by the missing vowels in Muts. In the vowels spreadsheet I changed the background color of these so they stand out. I'm not sure when Muts loses the vowel

Vowel *aw

far to ta taw Muts/Mao aw → a
o a a o Thok aw → o
mouth ka kāoš kaw Thok aw → a or o kou
fur ma me mao maw T/M/M aw → a/e/ao mao
forest tse tsao tsaw Could be *ew, but I'm too
head ta te tao taw seduced by that 'mao' tou
a e ao

The first of these sets can be distinguished from *o below by the fact that the Muts reflex is a. (This used to be a bigger set... I keep moving things out of it.) There's not much reason for *aw in particular, except that it's easy to get both o and a out of it.

The next set can also be distinguished from *o, whose Thok and Muts reflexes are both *o. The original must be more frontal; *ew is possible too. The Mandarin words all shout out that it's a diphthong.

Vowel *e

flax maṫi metsi matsi metsi Thok e → a
autumn nahupu nesufu nasumú nesubu exc. if there's an ^
sun, day, east (pa) hefa āma rabazhi šeba then it remains ê
west (phî) hefa?ii rababuhi šebahī
fire hefatsa amatsá rabasahi šebatse Mao e → a
hot (paṫû) hefatsi? āmatsi rabasi šebatsih
warm (kape) hexafa? āgamā ragapa šegabeh Zele e → a exc. in benir
right se á se but 'star' is odd in Thok and
cloud (pu) sepu āpú sepu Muts as well!
snake xata seh āš seš
all sesowo āzāwáo sezowo
south (khim) sexi āgio segim Thok 'hear' has weird î
happy (ku) hex āku šeku (yukuai, xingfu)
son te ta te
so that te? teh
bad an wen wanu van wen
hear, listen mîṫe mets matsīo mesea metsew
a e a a
cliff hemna aoná šemna e + consonant here
man xêma kemma kaomao 'husband' kemmew

guts ṫêmi tsemmi tsaomi tsemmi
star uni wejnih waniš benir weniš
ê eC ao

Strangely, there are no clear e/e/e/e correspondence sets, and no clear o/o/o/o. For awhile I thought of this one as "a with something that makes it e in Muts", with a mnemonic *ä; I only had *e in checked syllables (the second set here). Finally I realized that one or more of my correspondence sets must be the original *e, either this or the next one. It's a bit arbitrary, and the original sounds might have been distinguished many ways— e/ey is just a guess.

'son' and 'so that' are a minimal pair for something, though it's not clear what. :)

Vowel *ey

mountain ke kae key Thok, Zele ey → e
hill kejdi kaezí keydi Thok has some oddities
heavy xemîtû kaemiozī keymītü
steal kejpuns kaépyš keypuš Muts ey → e exc. before a stop
well, mine ṫwihi tssi saēšis 'mine' seyši
live ixe hexa aēkā reke šeykä Mao ey → ae
often ihehyô aešae rer šeysu
three ṫe te taé tey fat' may not be cognate
twelve tôkata taéga tey-kada 3'-'4'
egg ṫênwam taenūwaē teynuwan dan ?
compare ṫehâ tesa? taesá teysah Thok has a difference too so
then tse tsaē tsey it does seem diff. from e
fat tjems tsipaē tseyp- ey' is just a convenient guess
1s a ey
e e ae e

This may well be *e instead; Muts e → ae would parallel o → ao.

Vowel *i

love khîṫe ktsan gitšaē yizã gitzan
ear itta ?ita īzā ida
heart im īo šim xin
two ihwi īšys šišwe
pick ṫipô tsfuh tsipuš
light khikîtû kigiozi kigüdü
grass minam mwina mināo minam minam
tail mipa mifa mipā mipa
want nittû nit nizīo nidu-
green nime nin nināo (nã) ninen
pull (nyô) nizīo nidum ?
throat āowyš wiš
drink i lio yi
neck ṫîpu ( hy?uun ) tsikamá
small mimatû mi mi
wolf wîhî wiatsu wíatsū boabun wihatsü
i i i i
smell îhe īošy imši i + consonant
see îtû simti? iotí sheati simtih
some gimsuh giusūš neafuri gimsuh
rain mim mii mio mīm
water mîkhi mimki miogí meakihi mimki
frog tsimho tsiošwā tsimšV
wing ṫêpam tsišpāo tsišpam chibang
family miintsu mytsú measuhi mīntsu
bear child sintsih ytšiš sintzi
garden îčî wyty wintü (yin)dao
woman khîphî kiispy kispio kīspü
lead (vb) îtte iinda īndä
î iC io, y ea

These are the relatively friendly *i cognates, as well as *iC, discussed above under *m and *n.

Vowel *ī

nose phîṫi piitsi pitší pītzi
lion îthi siitsi sītsi
bird ṫîphî tsiifi tšímī tzībī
arm fêpam piipaw pípao pïmpam
back fêpu pípū pibuzh pïmpu
sheep mima miiwa mima mimazhi mīma
spring nihipa niisifa nizipá nīzipa
few pimô piiwu pimu pīmuh
daughter tii ti
left (dir) sii í
north ikhim siixi īgio shigimi sīgim
finger tsii tsi tsī (shou)zhi
sky twam tiihen tílae tīyen tian
sharp ṫe tsy tsï
do ne 'be' ni ni
i î ê ... ii i i

I couldn't find a good conditioning environment for Muts long ii, so I just punted it back to the protolanguage. It almost looks like it happened after unvoiced consonants, but that doesn't fit miiwa and kiispy, and it breaks down completely with uu below.

Vowel *o

fly kome gaomā home gomä
lie down, rest komô gwuh gaomūš gomuš Muts o → 0 / not sure when
sit mome mwun maomy` momun
summer nofa naomaē nobäy Mao o → ao
many pomî pwun paomy pomü
slow tome towa? taoma tomeh Muts 'cow' is weird
pea to?uu taowú towu prob. suffixed or s.t.
go tohe tos taozāo tsoshã tozam
hunt ṫokû tsox tsaokīo tsokü If Thok gets a^ zhuigan
tree ṫonam tsna tsaonāo tsonam then o → ê
branch ṫopi tsfi tsaopí tsopi (only 'elbow'; dunno
fear ohe joh yoše what's lost, maybe w!)
lentil tophî tfinh topinš
wrist wam so?on āowaē sowan wan
cow nona nynoh non?
sister ṫêṫam tsan tsatsae toza tsotsay jiejie
o o, 0 ao o
elbow ṫêṫu tsotsu tsaotsú tsowtsu ow = o in all but Thok zhou
throw ṫohî tssams tšaosāos tzosams
know ṫêṫâ ttsa? zaotšā toza dowtzah zhidao ?
sing ṫêṫamâ ttsawa? tatšāmá sumazhi totzámah
say ôtâ (?) psa? pazá pota podah
ê o, 0 ao, a o

Those ao in Mao were a big headache— they suggested a diphthong, and yet were clearly sometimes reflexes of *mC. Finally, I realized that I was looking too closely at Mao and that the simplest original was just *o.

As the sound changes column notes, when any of these words have a circumflex in Thok the vowel is ê rather than ô. But the circumflex usually indicates missing material, and what was lost here? Hard to say, but *ow at least gives something to lose, and relates *tsowtsu to Mandarin zhou.

Vowel *ū

who suu zu sū ? Thok: if ^ due to elision
root kunam guuna gunao gūnam then u → î genben
young uhu huuhu šūhu
pig tsuutsu tsú tsū Can't see any cond. factors zhu
earth, dirt uhî uuho?o wušāowá wūšowo so I assume ū is original
sea duu tuzh Goes to u in all but Muts.
tongue ôta suut útao sūtaw shetou
black uttû juuti? ludī yūdih Mao u → y after unvoiced
stone ṫuma tsuuwa tsuma tsūma (but also after 2 of 3 w!)
mushroom muhu muuhu mūšu mogu
ugly ṫînu tsuunv tsynu tsūšu
herd kuni kuunii kylí kūnyī
dog îhî wyš wuš
sow phîṫâ ptsun putsyš putsun bozhong
skin phînî puunnis pynis pūnnis ū + cons.
flow uunsa wysāo wūnsaw
snow čîmî tuunmam tymáo tūnmam
u uu u u

Again, I can't say where long u comes from in Muts— this seems like a pretty miscellaneous set of environments.

The Mao u/y alternation worried me a lot. It suggests umlaut, but that doesn't account for wysāo or tymáo. Here and in some other sets it looks like the conditioning factor is the voicelessness of the preceding consonant (except for tsuma).

Vowel *ey

winter ku?û kulú kuyu Generally u remains.
tie nuwī nuwi If Thok gets a^
blood ṫûṫi tsiitsi tsutsï then u → ô
bone ku gu gu gu
horn ku ku
priest kuhu kuhu Oddities:
temple kuha kūpas kupa Thok û in 3 words
crops kôhi kuhsi kus kussi Thok čîmê
cold kuṫyô kutšūlū kuzeyu kutzuh Muts û in 3 words
kick û yu Mao tones
deer uka ûxa lukā 'reindeer' yuka Mao y in 3 words lu
feather uma lumāo yumaw yumao
eye mu mu mu -
fish pumu pwu pumy` pumu
long psoh pusāoš pusaws
breast, chest pômî puhmun pušmyš pušmin
cut tsuhkin tšušky` tzuškin
split ṫômô tsuhma tšušmāo tzušmaw
foot ṫu tsu tsú tsu jiao
copper tsnohi tsunūi tsunohi
enemy tuts tutā tut-
2s tu tu
spit tôṫô tutsuh tútšūš tutzuš tutan
thin ume ūma umä
if ?wu? ūmú wumuh
sleep, dream upû sufuf ūmū fufub subub
blow une unu unu
louse sufimsin ūpīos supims-
hate uhâ suha? ūšá sušah
breath, wind uhiphi ūšipí roazh ? ušipi
never uhôhap ūúšae uhušay
freeze wûwi? wumí wumih
silver (fê)ṫuma zumao dumaw
nut, seed kuxî kxam kumC
mouse uhu sus 'rat' fush sus
night utta ûta uda
take tume twin tumy` tumi
u u u u
farm tsuntin tsyzyš tsuntu u + cons.
white čîmê tunmam tymao tunmäm
î uC yC

Pretty clear *u, except where it's not. I couldn't tell you where Muts gets these û or where the tones in Mao come from.

Vowel *ü

leg tu tu ty` tuy tui
and u y` ü
metal ṫumaxa tšimaetú tzümayt-
wash ûhe sis ís süs-
smooth ûhamyu isāmīo üsamyu
dig ahô ys wīs wüs
u i y i y
strong unî āonyš uwni
carry upe āopy` uwp?
horse uha hyha āoša šuwša
u y ao

'Leg' seems taken from Mandarin, which at least suggests that something more than *u is going on here, especially since *tu normally becomes u in Mao— compare the 2s pronoun tú, or tutā 'enemy'.

The next few words are a grab bag of inexplicables: something that goes to *u in Thok and *i or *y in the others. *ü is just a possibility.

The remaining words are another odd puzzle; I don't think there's enough data to really see what's going on.

Vowel *we

argue mwitâ mûsa? mutá mwetah Thok tyam lost it entirely
think mwîṫû mûtsi? mutsí mwetsih
sew nwihe nûs nusīo nwese Muts we → û
river ttwiku dûûxu zugú zwegu but u in hu?ytsim, kuu
clothes mwipa mûfa mupā mepazh mwepa and → 0 in 2 words
fox wiṫî hu?ytsim šwetsim (huli)
beech ṫwî tsy?uun tswehun ? Mao we → u
belly twî ?iih tylīš tweyī but → y after unvoiced duzi
stop twixe tûk tygīo tseyea tswegim
die wime hwun ymy remoa šwemo Zele we → e xiaomo
kill wihâ hûsam ysaoš šwesa shasi
spirit kwi kuu ky` kwe gui
knee twîtham tts tytsāo twetsam
seed tyam ?e tylāo tweyam
wi û u e

This, on the other hand, I really like. I originally had these as *ü too. Finally I noticed the common wi in Thok, and that it almost always paralleled û (unrounded u) in Muts. Deducing the original is another story. An initial *w is a no-brainer; but what's the vowel after it? *e is suggested by Zele and by Mandarin gui (pronounced guei).

Of course, how is it we have *wi but very little *we, *wa, *wo? Maybe we do, but the patterns just weren't as striking.

Vowel *yu

month, moon kyumi kyywi kiomi kyumi Muts yu → y
face myumyu mywyh miomīoš myumyus 'month' is oddly yy mian
give nysa? niotá nyuta
call pyutâ piotá pyutá Mao yu → io
play sysi? īosí syusih finally, → ī
girl tyukam tiokāo tyukam
fast tyume tywa? tioma tyumeh
near tyu ty tyü
fog, mist sysimma iomā syusimma
bug yuhi sysi iozīo syuzü
come tsun tiosy` tyusun
yu y io

For this set, *yu roughly shoulders aside most other possibilities, thanks to the reflex io in Mao. This is phonetically /iw/, but it's quite plausible that the spelling reflects an earlier pronunciation.

Again, it's a little odd not to see clear parallels *ya, *ye, etc.