User:Soap/East

The parent language. Had only /h/ using internal reconstruction, but this could have been several fricatives that merged.

Bilabials:       p    b    m
Alveolars:       t    d    n    l    r
Postalveolars:   č    ǯ
Palatals:                       (y)
Velars:          k    ġ    ŋ    (Ø)       h

Vowels were /a i u ə/.

The syllable structure is CVC, with final consonants restricted to /t n l r č k ġ ŋ/; the "g" was probably [ɣ] in this position and might have actually arisen from earlier /h/ or from the fricatives that themselves led to /h/. Underlying root-final voiced stops are possible, as is /-p/, but they were not realized as such.

Notes on diachronics

Ideally, User:Soap/Birch is a branch of this which splits off after the lenition of /č ǯ/ to /s y/ (conditionally) and probably also after a re-shaping of the vowel system to /a e i o u/ but before most of the other shifts that make Birch what it is.

Motherport branch

Ideally, as above, this is closely related to Commercial/Birch but without its most characteristic shifts. Thus the shifts shared with Birch are placed as high up as possible. This is a sonorant-rich branch, ideally with a nasal for every stop, but unlike Birch it begins with no palatals and the only postalveolars are /č ǯ/ with no nasal member.

1. The vowels i u shifted to e o in closed syllables except those ending with one of /r ŋ/. If /e o/ later shift back to /i u/, this shift could even be skipped, but they must have been at least allophonically different in order to trigger uvularization and palatalization.
2. The voiced stops b g shifted to w ɣ in initial position (in Birch they shifted to /w Ø/ everywhere, but this may have been stepped). Initial gʷ shifted to w as well. Note that Birch shifted the labiovelars to labials early on, which means this branch might need to as well, which would remove them from the language.
3. The voiceless velar stops k kʷ shifted to q qʷ before any /e a o/.
4. Remaining g shifted to k, even in situations where the expected allophone was [ɣ]. This saves the language from having more /q/ than /k/.
5. Possibly tl tr shifted to ʈ (retroflex). (There is no convenient notation for this in the usual writing system I use.)
6. Maybe čd shifted to ʈʈ (retroflex geminate); This cluster would have been rare however.
7. In word-initial position before a schwa, the alveolars t d n l shifted to ʈ ɖ ɳ ɭ. A retroflex flap /ɽ/ is unlikely; however it is a very likely allophone of the voiced stop at a later stage of the language.

   Note that both Dravidian and Australian lean towards not having retroflexes word-initially, but that some of this may be due to their originating from consonant clusters.

8. The sequences əi əu shifted to i u.
9. The vowels e ə o shifted to i a u unconditionally. This means that /qi/ exists, but its surface realization remained near [qe]. There was no phonemic palatalization before /i/, but since /ki ke/ > /ki qi/, it can still arise even after this shift.

   NOTE: It may be possible to write this timeline without using /e o/.

A daughter branch of this might shift /k q/ > /ć k/ and then /ć > ṭ/ like in Birch, but with different conditions, so they would only overlap rather than matching.