Phrase evolution game

[Nortaneous]

蒹葭苍苍
白露為霜
所渭伊人
在水一方
溯洄从之
道阻且長
溯游从之
宛在水中央

Kem tra chen chen
Biak pias jü chü
Sha jüs tsü mü
Jü sytsur tsit pü

Sypias jü dzorn ty
Chü cha cha nrytrarn
Sypuos nrydru dzwo ty
Twor jüt sytsur chü tsü

[kʰɛm ʈʰɑ c͡ɕʰen c͡ɕʰen
bjæk pʰjæs ɟ͡ʑᶣy c͡ɕʰᶣy
ɕæ ɟ͡ʑys t͡sʰy mᶣy
ɟ͡ʑᶣy sɨt͡sur t͡sʰit pᶣy

sɨpʲæs ɟ͡ʑᶣy d͡zɔɳ tʰɨ
c͡ɕᶣy c͡ɕʰæ c͡ɕʰæ ɳɨʈɑɳ
sɨpʷos ɳɨɖu dzʷo tʰɨ
tʰʷor ɟ͡ʑᶣyt sɨt͡sur c͡ɕᶣy t͡sᶣy]

1) Retroflex finals eject rhotic coarticulation onto the vowel, which is then reinterpreted as a coda -r
2) Coda -r > infixed -l- (this happened in a Mandarin dialect)
3) Consonantal coarticulations affect preinitial ɨ
4) Palatals > tɕ dʑ (mostly notational)
5) 之 develops into an affix, taking an epenthetic copy vowel after a final consonant and taking the form -t otherwise
6) Epenthetic copy vowels also develop to break up Cl clusters (attested in some Papuan languages)
7) Cʷo > Cuo, labiopalatalization (only occurring before y?) is lost
8) Pj > Ptɕ
9) e > ie

khem ṭha cien cien
bjäk pchäs jü chü
śä cüs tshü mu
cü sïtsulu tshit pü

sipcäs jü dzolónot
cü chä chä ṇïṭalan
sïpuos ṇïḍu dzuot
thuoluo jüt sïtsulu cü tsü

kʰɛm ʈʰɑ tɕien t͡ɕien
bdʑæk ptɕʰæs dʑy tɕʰy
ɕæ tɕys tsʰy my
tɕy sɨtsuˈlu tsʰit py

siˈptɕæs dʑy dzɔˈlɔnɔt
tɕy tɕʰæ tɕʰæ ɳɨʈɑˈlɑn
sɨˈpuos ɳɨˈɖu dzuot
tʰuoˈluo dʑyt sɨt͡suˈlu tɕy tsy

[Rounin Ryuuji]

khem ṭha cien cien
bjäk pchäs jü chü
śä cüs tshü mu
cü sïtsulu tshit pü

sipcäs jü dzolónot
cü chä chä ṇïṭalan
sïpuos ṇïḍu dzuot
thuoluo jüt sïtsulu cü tsü

[kʰɛm ʈʰɑ tɕien t͡ɕien
bdʑæk ptɕʰæs dʑy tɕʰy
ɕæ tɕys tsʰy my
tɕy sɨtsuˈlu tsʰit py

siˈptɕæs dʑy dzɔˈlɔnɔt
tɕy tɕʰæ tɕʰæ ɳɨʈɑˈlɑn
sɨˈpuos ɳɨˈɖu dzuot
tʰuoˈluo dʑyt sɨt͡suˈlu tɕy tsy]

(1.) Reduplicated forms fuse into a single word, often with voicing of the medial consonant; the first element is usually stressed, usually accompanied by a reduction of the second syllable, [ie] > [i].
(1.1) 人 and 方 become affixes
(1.2) Reduction of the vowels in affixed syllables, [ɔ] > [ə], [y] > [ɨ]
(1.3) Second epinthesis, inserting ə after [p b] followed by a palatal affricate
(2) Breaking of [æ ɛ] > [ɛə], [y] > [ɛʊ]
(3) Retroflex stops become affricates [ʈʰ ɖ] > [ʈ͡ʂʰ ɖ͡ʐ];
(3.1) Palatal affricates eject [j] into the following syllable if the next vowel is not [i]
(3.2) [jɛə] >[iə], [jɛʊ] > [jey]
(3.3) Some forms are reduplicated to avoid homophony, writers in the language invent spaces.
(3.4) Respelling of some reduplicated forms with 々
(4) The linguists documenting the language do not agree on Romanisation or Notation.

蒹 葭 苍々
白 露露 為 霜
所 渭 伊人
在々 水 一方

溯 洄 从之
道 阻々 且 長
溯 游 从之
宛 在々 水 中 央

Keom cha chienjin
bëjiek pëch'iós jieu ch'ieu
Shië chieus ts'umï
chieujië sïtsulú ts'itëbï

Shi pëchiós jieu dzolonët
Chieu ch'ieujië ch'ië nrïchalán
sïpuos nïdzu dzuët
t'uolu chieujië sïtsulú chieu tseu

[kʰɛəm ʈ͡ʂʰɑ c͡ɕienɟ͡ʑin
bədʑiək pə'tɕʰiəs ɟʑjey tɕʰjey
ɕiə tɕjeys tsʰymɨ
tɕjeyɟʑiə sɨtsuˈlu tsʰitəbɨ

ɕi.pə'tɕiəs dʑjey dzɔˈlɔnət
tɕjey tɕʰiəɟʑiə tɕʰiə ɳɨʈ͡ʂʰɑˈlɑn
sɨˈpuos ɳɨˈɖ͡ʐu dzuət
tʰuoˈluo tɕjeyɟʑiə sɨt͡suˈlu tɕjey tsɛʊ]

[bradrn]

[kʰɛəm ʈ͡ʂʰɑ c͡ɕienɟ͡ʑin
bədʑiək pə'tɕʰiəs ɟʑjey tɕʰjey
ɕiə tɕjeys tsʰymɨ
tɕjeyɟʑiə sɨtsuˈlu tsʰitəbɨ

ɕi.pə'tɕiəs dʑjey dzɔˈlɔnət
tɕjey tɕʰiəɟʑiə tɕʰiə ɳɨʈ͡ʂʰɑˈlɑn
sɨˈpuos ɳɨˈɖ͡ʐu dzuət
tʰuoˈluo tɕjeyɟʑiə sɨt͡suˈlu tɕjey tsɛʊ]

1. ɛə → ɛjə, ɛʊ → ɛwʊ, ey → ewy, uo → uə, ie → iə
2. Middle vowel of trisyllables deletes
3. Retroflexes and palatals merge to palato-alveolars (except for /ɳ/)
4. Coda obstruents delete
5. /j/ deletes after /tʃ/ /dʒ/
6. Aspirates and voiceless consonants merge
7. Stress becomes regularly word-final
8. Linguists give up, create new romanisation from scratch

keʼyəm ca cienjin
bəjie pəcie jewʉ cewʉ
xie cewʉ tsʉmɨ
cewjie sɨlu tsibɨ

xicie jewʉ dzoʼlnə
cewʉ ciejie cie rnɨlan
sɨpue rnɨju dzue
tuelue cewjie sɨlu cewʉ tseʼwuʼ

[kɛ.jəm tʃɑ tʃiən.dʒin
bə.dʒiə pə.tʃiə dʒe.wy tʃe.wy
ʃiə tʃe.wy tsy.mɨ
tʃew.dʒiə sɨ.lu tsi.bɨ

ʃi.tʃiə dʒe.wy dzɔ.lnə
tʃe.wy tʃiə.dʒiə tʃiə ɳɨ.lɑn
sɨ.puə ɳɨ.dʒu dzuə
tuə.luə tʃew.dʒiə sɨ.lu tʃe.wy tsɛ.wʊ]

Hmm, which branch of Sino-Tibetan do you think we’re in now?

[Rounin Ryuuji]

Hmm, which branch of Sino-Tibetan do you think we’re in now?

Time for World War II?

ケヨン チヤ チヨウンヂン
ボジヨウ ポチヨウ デユ テユ
シヨウ テユ チユミ
シヨウジヨウ シル チビ

シチ デユ ゾノ
テユ チヂ チョウ イラン
シポ イジユ ズヲ
ツヲロ チヨウヂ ルテ ユセユ

[kɛ.jon tʃɑ c͡ɕoːn.ɟ͡ʑin
bo.ɟ͡ʑoː po.c͡ɕoː de.jy te.jy
ɕoː te.jy c͡ɕy.mi
c͡ɕoː.ɟ͡ʑoː ɕi.lu c͡ɕi.bi

ɕi.c͡ɕi de.jy zo.no
te.jy c͡ɕi..ɟ͡ʑi c͡ɕoː i.lɑn
ɕi.po i.ɟ͡ʑy zuwo
two.lo c͡ɕoː.ɟ͡ʑi ɕi.lu te.jy se.jy]

Keyon cha chounjin
bojou pochou deyu teyu
shou teyu chumi
shoujou shilu chibi

Shichi deyu zono
Teyu chiji chou ilan
shipo iju zuwo
twolo chouji shilu teyu seyu

(1) Sequences [ji wu wy] lose initial glides, however [e i] eject [j] before another vowel except [i], and [y u o] eject [w] before all but [u]
(1.1) [jʊ] > [jy], [uə wə] > [wo] (but preserving ejected [w] after [u y]), [ə ɔ] > [o], [ɨ iə] > [i]), [e ɛ] merge
(1.2) [lw] > [l]
(2) Before [i], [ts tʃ] > [c͡ɕ], [dz dʒ] > [ɟ͡ʑ], [s ʃ] > [ɕ], [ts dz] elsewhere become [s z]
(2.1) Before [e], [ʃ tʃ dʒ] depalatalise to [s t d], blocked by [w] after [e]
(2.2) [ew, io] > [joː], loss of [j] after a palatals
(3.) Loss of preconsonantal [l], and intial [ɳ]; [m] > [n] terminally
(4.) Japanese invasion (and presumably occupation), change of orthography to Katakana.

Edit: Forgot to add new Romanisation, based on Hepburn.

[Otto Kretschmer]

@Otto Kretschmer

Can we add texts of our own to be evolved, or would you like to be the sole arbiter of that duty? I have a particular text I think would be interesting for the next round.

Of course you can! I will add next one in PIE and after that you are free.

I just had few texts in mind and Iike to see past languages evolvingv diffrrently. But modern languages are equally valid

PS Old Chinese sounded terrible to me with all those braks praks and glottal stops

UDHR in PIE (no idea if translation is correct)

h₂ólyoes ǵʰmónes h₁léwdʰeroes somHóeskʷe gʷr̥Htóteh₂ti h₃r̥ǵtúsukʷe ǵn̥h₁yóntor. éybʰos dh₃tóy ménos ḱḗrkʷe h₁stés h₂énteroeykʷe sm̥h₂éleyes bʰréh₂tr̥bʰos swé h₂éǵoyh₁n̥t.

[bradrn]

[kɛ.jon tʃɑ c͡ɕoːn.ɟ͡ʑin
bo.ɟ͡ʑoː po.c͡ɕoː de.jy te.jy
ɕoː te.jy c͡ɕy.mi
c͡ɕoː.ɟ͡ʑoː ɕi.lu c͡ɕi.bi

ɕi.c͡ɕi de.jy zo.no
te.jy c͡ɕi..ɟ͡ʑi c͡ɕoː i.lɑn
ɕi.po i.ɟ͡ʑy zuwo
two.lo c͡ɕoː.ɟ͡ʑi ɕi.lu te.jy se.jy]

1. Palatals become retroflex except before /i y/
2. Unstressed short vowels delete; unstressed long vowels become /ə/
3. /tj dj sj kj/ → /tɕ dʑ ɕ tɕ/
4. Linguists get hold of this language and define a horrible Pinyinish romanization for it

qon cha chenjin
bzhoo pchoo jü qü
shoo qü qmi
chezhoo xlu qbi

xqi jü ssno
qü qji choo lan
xpo jü sswo
twlo cheji xlu qü xü

[tɕon tʂɑ tʂə.ndʑin
bdʐoː ptʂoː dʑy tɕy
ʂoː tɕy tɕmi
tʂə.dʐoː ɕlu tɕbi

ɕtɕi dʑy zno
tɕy tɕdʑi tʂoː lɑn
ɕpo dʑy zwo
twlo tʂə.dʑi ɕlu tɕy ɕy]

(Huh, looks like Tibetan Mandarin…)

PS Old Chinese sounded terrible to me with all those braks praks and glottal stops

Interesting, I quite like that æsthetic.

UDHR in PIE (no idea if translation is correct)

h₂ólyoes ǵʰmónes h₁léwdʰeroes somHóeskʷe gʷr̥Htóteh₂ti h₃r̥ǵtúsukʷe ǵn̥h₁yóntor. éybʰos dh₃tóy ménos ḱḗrkʷe h₁stés h₂énteroeykʷe sm̥h₂éleyes bʰréh₂tr̥bʰos swé h₂éǵoyh₁n̥t.

I think I’ll pass on this one, at least until someone else can get rid of all that horrible PIE-ist /h₁ h₂ h₃/ notation (which, admittedly, should happen pretty quickly).

[Otto Kretschmer]

I like the way you guys changed the Chinese text.

[Rounin Ryuuji]

qon cha chenjin
bzhoo pchoo jü qü
shoo qü qmi
chezhoo xlu qbi

xqi jü ssno
qü qji choo lan
xpo jü sswo
twlo cheji xlu qü xü

[tɕon tʂɑ tʂə.ndʑin
bdʐoː ptʂoː dʑy tɕy
ʂoː tɕy tɕmi
tʂə.dʐoː ɕlu tɕbi

ɕtɕi dʑy zno
tɕy tɕdʑi tʂoː lɑn
ɕpo dʑy zwo
twlo tʂə.dʑi ɕlu tɕy ɕy]

[son tʂɑ tʂə.nʑin
dʐou tʂou dʑjɛʊ ɕjɛʊ
ʂou ɕjɛʊ swi
dʐou ɕjɛʊ c͡ɕwi

tɕei ʑjɛʊ ʐou
tɕiw dʑiw tʂou lɑn
ʂwɛʊ dʑiw zwɛʊ
tʂwɛʊ dʑei ɕjɛʊ ɕjɛʊ ɕjɛʊ]

Son cha chenzhin
Jou chou jieu shieu
Shou shieu sui
Jou shieu chui

Chei zhieu zhou
Chiu jiu chou lan
Shueu jiu zweu
Chueu jei shieu shieu shieu.

(1) Deletion of initial [p b ɕ] when preceding another consonant, unless the following consonant is a palatal, at which point the initial is geminated, and [ɕl] > [ɕj], [cɕp cɕm] > [cɕw]
(1.1) Deletion of syllables in [ə].
(2) Depalatalisation and disaffrication of palatal affricates, depalatalisation blocked by following [i j y], both blocked by gemination;
(2.1) Following [j] and gemination disappear, loss of gemination triggers compensatory lengthening
(2.2) Where [j] is lost before [u], the [u] has become [y]
(2.3) [zn] > [ʐ], [twl] > [tlw] > [tɕw]
(2.4) [ɕw] > [ʂw]
(3.) [y wo] > [ɛʊ wɛʊ], [yː] > [iw], [oː o] > [ou] (blocked by coda), [iː] > [ei]
(4.) Independence from the mainland established, a new Roman orthography developed for official use.

I quite like this. Now a part of me wants to make a Sinitic language to go with that Japonic one I've been working on.

Edit: More BBCode fixes (it and I do not get along).

[Otto Kretschmer]

Is it feasible that a Sinitic lang would go through this much change and remain monosyllabic? In two verses number of syllabes actually went down ftom 5 to 3

[dɮ the phoneme]

h₂ólyoes ǵʰmónes h₁léwdʰeroes somHóeskʷe gʷr̥Htóteh₂ti h₃r̥ǵtúsukʷe ǵn̥h₁yóntor. éybʰos dh₃tóy ménos ḱḗrkʷe h₁stés h₂énteroeykʷe sm̥h₂éleyes bʰréh₂tr̥bʰos swé h₂éǵoyh₁n̥t.

I know next to nothing about early IE diachronics, or what would really be plausible here, so let's just take this somewhere unique.

1. labiovelars merge with labials: kʷ gʷ > p b
2. satemization: ḱ ǵ > ś ź
3. {o e} {ō ē} > a ā adjacent to h₂, > o ō adjacent to h₃
4. laryngeals merge (to something like ɦ) in coda position
5. tonogenesis: VH > V̀, V́H > V̂
6. laryngeal loss: larynges not adjacent to a vowel > a, other laryngeals disappear, except h₁ or h₂ syllable-initially before a vowel, which are retained as h
7. tone split: Dʰ(C)V Dʰ(C)V̀ Dʰ(C)V́ Dʰ(C)V̂> T(C)V̀ T(C)V̀ T(C)V̌ Dʰ(C)V᷈

ˈhólyoes ˈšmǒnes ˈléwtèroes somˈhóespe braˈtótàti aržˈtúsupe žnaˈyóntor. ˈéypòs daˈtóy ˈménos ˈśḗrpe aˈstés ˈhánteroeype smáleyes ˈbra᷈tr̥pòs ˈswé háźoyant.

[Otto Kretschmer]

In 2 days we have a toneless Sinitic language and a tonal IE one.. nice!

[Rounin Ryuuji]

Is it feasible that a Sinitic lang would go through this much change and remain monosyllabic? In two verses number of syllabes actually went down ftom 5 to 3

Well, it went through stages of being polysyllabic, and then back to being monosyllabic. I'm sure it would've changed more syntactically — this is just the modern reading of an old poem with a few very minor changes over the years. It probably sounds Chaucerian to the modern-day speakers.

[Nortaneous]

1. labiovelars merge with labials: kʷ gʷ > p b
2. satemization: ḱ ǵ > ś ź
3. {o e} {ō ē} > a ā adjacent to h₂, > o ō adjacent to h₃
4. laryngeals merge (to something like ɦ) in coda position
5. tonogenesis: VH > V̀, V́H > V̂
6. laryngeal loss: larynges not adjacent to a vowel > a, other laryngeals disappear, except h₁ or h₂ syllable-initially before a vowel, which are retained as h
7. tone split: Dʰ(C)V Dʰ(C)V̀ Dʰ(C)V́ Dʰ(C)V̂> T(C)V̀ T(C)V̀ T(C)V̌ Dʰ(C)V᷈

ˈhólyoes ˈšmǒnes ˈléwtèroes somˈhóespe braˈtótàti aržˈtúsupe žnaˈyóntor. ˈéypòs daˈtóy ˈménos ˈśḗrpe aˈstés ˈhánteroeype smáleyes ˈbra᷈tr̥pòs ˈswé háźoyant.

1) h > 0
2) oe > oː
3) o > a
4) long high > long falling
5) contrastive vowel length is lost
6) remaining R̥ > oR
7) contrastive stress is lost

ályos šmǒnes léwtèros samóspe bratátàti aržtúsupe žnayántar. éypàs datáy ménas śêrpe astés anteroype smáleyes bra᷈torpàs swé áźayant.

would mess with tone and restore vowel length but i don't know enough about tone co-occurrence restrictions to make that work

[Rounin Ryuuji]

1) h > 0
2) oe > oː
3) o > a
4) long high > long falling
5) contrastive vowel length is lost
6) remaining R̥ > oR
7) contrastive stress is lost

ályos šmǒnes léwtèros samóspe bratátàti aržtúsupe žnayántar. éypàs datáy ménas śêrpe astés anteroype smáleyes bra᷈torpàs swé áźayant.

would mess with tone and restore vowel length but i don't know enough about tone co-occurrence restrictions to make that work

(1) Preceding vowels undergo [i] or [u] mutation, also conditioned by [j w].
(1.1) Insertion of epinthetic [i] after palatals, or [ɨ] elsewhere, to break up certain now-illegal clusters.
(2) Replacement of tonal accent with a stress accent (final heaviest syllable in a word becomes stressed; if all syllables are light, stress falls upon the penult, which is lengthened).
(2.1) Loss of overshort vowels;
(2.2) Terminal nasal vowels are lost, with nasalisation and lengthening of rpeceding vowels.
(3.) Contraction of vowel-semivowel and vowel-semivowel-vowel sequences to long vowels, [eja] > [eː], [ije ej] > [iː], [ew] > [joː], [øj] > [jyː]; nasalisation blocks contractions, but [ejãː] > [jãː]
(4.) Attestation, /s/ is usually apical [s̺] except before /t/, where it is dental [s̪], /t d n/ are dental [t̪ d̪ n̪]. Romanisation altered somewhat for ease of enterability.

Eljos shĩnes ljôteros samôsype bratetêti orzhitusûpe zhinjãtar. Êpas datê menas shêrype astes ãterjûpe symalîs bratorpas swe azhjãt.
['el̪.jos̺ 'ɕĩː.n̪es̺ 'ʎjoː.t̪e.ros̺ s̺a'moːs̺ɨ.pe bra.t̪e't̪eː.t̪i or.ʑi.tu'suː.pe ʑin'jãː.t̪ar || 'eː.pas̺ d̪a't̪eː me'n̪as̺ ɕeː.rɨ.pe 'as̪.t̪es̺ ãː.t̪e'rjyː.pe s̺ɨ.ma'ʎiːs̺ bra't̪or.pas̺ s̺we a'ʑjãt̪]

[bradrn]

Eljos shĩnes ljôteros samôsype bratetêti orzhitusûpe zhinjãtar. Êpas datê menas shêrype astes ãterjûpe symalîs bratorpas swe azhjãt.
['el̪.jos̺ 'ɕĩː.n̪es̺ 'ʎjoː.t̪e.ros̺ s̺a'moːs̺ɨ.pe bra.t̪e't̪eː.t̪i or.ʑi.tu'suː.pe ʑin'jãː.t̪ar || 'eː.pas̺ d̪a't̪eː me'n̪as̺ ɕeː.rɨ.pe 'as̪.t̪es̺ ãː.t̪e'rjyː.pe s̺ɨ.ma'ʎiːs̺ bra't̪or.pas̺ s̺we a'ʑjãt̪]

1. ʎ(j) → j
2. y → i
3. Lenition of intervocalic consonants: stops voice (in unstressed syllables only), voiced stops fricativise, and fricatives voice
4. Cj → Cː across a syllable boundary
5. Vowel nasalisation ejected onto the previous consonant, or Ṽ → ŋV if there is no previous consonant

Ellos ñīnes jōderos samōzybe bradetēdi orzhiduzūbe zhinnādar. Ēbas datē menas shērybe astes ŋāderrībe symajīs bratorpas swe azhñat.
['el̪.l̪os̺ 'ɲiː.n̪es̺ 'joː.ðe.ros̺ s̺a'moːz̺ɨ.βe bra.ðe't̪eː.ði or.ʑi.ðu'zuː.βe ʑin'naː.ðar || 'eː.βas̺ d̪a't̪eː me'n̪as̺ ɕeː.rɨ.βe 'as̪.t̪es̺ ŋaː.ðer'riː.βe s̺ɨ.ma'jiːs̺ bra't̪or.pas̺ s̺we aʑ'ɲat̪]

EDIT: As an experiment, I also tried running numbers 1–5 through: /h₁oinos dwéh₃ tréyes kʷetwóres pénkʷe/ → /hoinos dwô tréyes petwóres pénpe/ → /oinas dwâ tréyes petwáres pénpe/ → /ˈjyːn̪as ˈd̪eː ˈt̪reːs̺ pet̪'wares̺ 'pẽːpe/ → /ˈjiːn̪as ˈd̪eː ˈt̪reːs̺ pet̪'wares̺ 'meːβe/ jīnas dē trēs petwares mēbe. Please double-check my work if you want!

[Rounin Ryuuji]

Ellos ñīnes jōderos samōzybe bradetēdi orzhiduzūbe zhinnādar. Ēbas datē menas shērybe astes ŋāderrībe symajīs bratorpas swe azhñat.
['el̪.l̪os̺ 'ɲiː.n̪es̺ 'joː.ðe.ros̺ s̺a'moːz̺ɨ.βe bra.ðe't̪eː.ði or.ʑi.ðu'zuː.βe ʑin'naː.ðar || 'eː.βas̺ d̪a't̪eː me'n̪as̺ ɕeː.rɨ.βe 'as̪.t̪es̺ ŋaː.ðer'riː.βe s̺ɨ.ma'jiːs̺ bra't̪or.pas̺ s̺we aʑ'ɲat̪]

jīnas dē trēs petwares mēbe
/ˈjiːn̪as ˈd̪eː ˈt̪reːs̺ pet̪'wares̺ 'meːβe/

(1) Initial [ŋ ɲ] > [g j], [ji] > [i]
(1.1) [ʑj] > [ʑʑ] > [ɕɕ]
(1.2) [ð β] merge to [β], then [β] > [w]
(1.3) [ji wu] > [i u]
(1.4) [aiː] > [eːi]
(2) Quantitative metathesis of vowel length from initial syllable to a medial, except that [ɨ] blocks the change; metathesis from second syllable to first in disyllables, carrying stress with it
(2.1) Between or terminally following long vowels, [e] > [ɨ], inflectional syllable [es] > [ɨs];
(2.2) The unstressed word [s̺we] > [s̺wɨ]
(2.3) Contractions of vowel-semivowel-vowel sequences: [ɨwi] > [ɨu], [ɨwe] > [ɛʊ], [iu] > [iy], all three later merge to [ɛʏ].
(3.) Reduction of unstressed [a o] > [ə]

Elləs înys jəwêrəs səmôzeu brəwêteu ərzheuzûwy zhinnâwər. Êwəs dâty mynas shêreu astys gâwyrrîwy symêis brətorpəs swy əsshat.
['el̪.l̪əs̺ 'iː.n̪ɨs̺ 'jə.we.rəs̺ s̺ə'moːz̺ɛʏ brə.'weːt̪ɛʏ ər.ʑɛʏ'zuː.wɨ ʑin'naː.wər || 'eː.wəs̺ 'd̪aː.t̪ɨ mɨ'n̪as̺ ɕêː.rɛʏ 'as̪.t̪ɨs̺ gaː.wɨr'riː.wɨ s̺ɨ.meːs̺ brə't̪or.pəs̺ s̺wɨ əɕ'ɕat̪]

Do we have some sort of Old Notwelsh here?

[Emily]

As an experiment, I also tried running numbers 1–5 through: /h₁oinos dwéh₃ tréyes kʷetwóres pénkʷe/ → /hoinos dwô tréyes petwóres pénpe/ → /oinas dwâ tréyes petwáres pénpe/ → /ˈjyːn̪as ˈd̪eː ˈt̪reːs̺ pet̪'wares̺ 'pẽːpe/ → /ˈjiːn̪as ˈd̪eː ˈt̪reːs̺ pet̪'wares̺ 'meːβe/ jīnas dē trēs petwares mēbe. Please double-check my work if you want!

Applying the subsequent changes to these gets us:

înəs dê três petwares meu
[ˈiːn̪əs ˈd̪eː ˈt̪reːs̺ pet̪'wares̺ 'mɛʏ]

Elləs înys jəwêrəs səmôzeu brəwêteu ərzheuzûwy zhinnâwər. Êwəs dâty mynas shêreu astys gâwyrrîwy symêis brətorpəs swy əsshat.
['el̪.l̪əs̺ 'iː.n̪ɨs̺ 'jə.we.rəs̺ s̺ə'moːz̺ɛʏ brə.'weːt̪ɛʏ ər.ʑɛʏ'zuː.wɨ ʑin'naː.wər || 'eː.wəs̺ 'd̪aː.t̪ɨ mɨ'n̪as̺ ɕêː.rɛʏ 'as̪.t̪ɨs̺ gaː.wɨr'riː.wɨ s̺ɨ.meːs̺ brə't̪or.pəs̺ s̺wɨ əɕ'ɕat̪]

1a. Geminates fortify in all intra-word positions; single consonants (or first consonant in cluster) fortify in onset of stressed syllable: l̪l̪ > ð, rr > z, nn > d, ɕɕ > t͡ɕ; d > dʰ > d͡z, d̪ > d̪ʰ > d̪͡z̪, t̪ > t̪ʰ > t̪͡s̪, z > d, j > ʒ, w > β, m > b, n̪ > d̪
1b. Consonants lenite after a vowel if either word-final or followed by another consonant: s̺, s > h > ∅; r > ∅; t̪ > s̪
2. Palatalization: consonants (except labials and palatals) become palatals if followed by i, e, ɛʏ, ɨ, or if preceded by i: r > j; n̪ > ɲ; d > ɟ; t̪ > c; s̺ > ʃ; z̺ > ʒ; d̪͡z̪, d͡z > d͡ʒ; t̪͡s̪r > t͡ʃj
3a. Distinctive vowel length lost; vowels are long in stressed open syllables, short otherwise (this change remains productive throughout all following changes)
3b. ə, ɨ > a, i if stressed, if CC_C or C_CC, or if the word is a monosyllable; deleted otherwise
3c. Vw, Vj > Vu, Vi unless followed by another vowel (uw, ij > u, i); ɛʏ merges into eu
3d. Unstressed diphthongs become monophthongs: eu > œ, au > ɔ
4. Dentals become alveolars; alveolars (and apicals) become retroflex; alveo-palatals merge with palatals: d > ɖ;,s̺ > ʂ; s̪ > s; ð > z, d̪ > d, t̪͡s̪ > t͡s, d̪͡z̪ > d͡z; ɕ > ʃ, ʑ > ʒ, t͡ɕ > t͡ʃ
5. Regressive voicing in obstruent clusters

Ez iń źawer ẓboźeu bravet́eu źeuḍu źidźau.
['ez 'iɲ 'ʒaːwer 'ʐboːʒœ bra'βeːcœ ʒœ'ɖuː ʒi'd͡ʒau]

Eu dzat́ mda śejeu at́ gaud́iu śme bratsop ṣwi tśas.
['eu 'd͡zac 'mdaː ʃejœ 'ac gɔ'ɟiu ʃme bra't͡sop ʂwi 't͡ʃas]

Iń dźe tśje pezvaje beu.
[ˈiɲ ˈd͡ʒeː ˈt͡ʃjeː pez'βaːje 'beu]

[Emily]

I'd like to add in the rest of the PIE numbers (swéḱs septḿ̥ h₃eḱtṓw h₁néwn̥ déḱm̥t ḱm̥tóm) but I'm having trouble wrapping my head around the tone stuff in the first couple of posts tbh

[dɮ the phoneme]

I'd like to add in the rest of the PIE numbers (swéḱs septḿ̥ h₃eḱtṓw h₁néwn̥ déḱm̥t ḱm̥tóm) but I'm having trouble wrapping my head around the tone stuff in the first couple of posts tbh

As per mine and Nort's changes, *swéḱs *septḿ̥ *h₃eḱtṓw *h₁néwn̥ *déḱm̥t *ḱm̥tóm should become swéś(s) septóm aśtôw néwon déśomt śomtóm. Ronin Ryuuji seems to have decided to just drop tone altogether though, so it seems like you don't need to worry about it.

[Karch]

1. diphthongs monophthongize even in stressed syllables: au eu iu ou > ɔ: œ: y: o:
2. unstressed vowels reduce: a œ e i ɔ o u > ə ə ɪ ɪ ə ʊ ʊ
3. front rounded vowels unround
4. consonant clusters in unstressed syllables reduce, dropping the less sonorous segment
5. final voiced consonants (with the exception of nasals and approximants) gain an epenthetic schwa
6. retroflexes merge with postalveolars, except before consonants, where they become alveolars
7. long vowel breaking: e: o: a: > ie uo oa
(7b. Cji Cwu > Ci Cu)
8. s drops word-finally, making vowel length phonemic

Ĕza iñ žawir zboža baveḱa žaǯu žiǯo̬. E̬ ʒaḱ mda šeja aḱ gaǵi šme bacop swi čă.
['ezə 'iɲ 'ʒoawɪr 'zbuoʒə bə'βiecə ʒə'd͡ʒuː ʒɪ'd͡ʒɔː || 'ɛː 'd͡zac 'mdoa 'ʃejə 'ac gə'ɟiː ʃme bə't͡sop swɪ 't͡ʃa]
iñ, ǯe, če, pizvaji, be̬
[ˈiɲ ˈd͡ʒie ˈt͡ʃie pɪz'βoajɪ 'bɛː]