So, here's the charts best as I can describe them:
Phonotactics:
Word-initial consonants: p'/t'/k'/pʰ/tʰ/kʰ/p/t/k/ts/f/s/m/l/j/w
Word-final consonants: pʰ/tʰ/kʰ/p/t/k/n/r
/pʰ, tʰ, kʰ/ > /ɸ, s, x/ word-finally if the next word begins with any consonant (/läpʰ/ > /läɸ/) with the exception of /f, s, n, m/ where they merely de-aspirate (/pʰekʰ/+/ɸän/ > /pʰekɸän/). There is also greater variation for /tʰ/ which can appear as /θ~ð/ for some speakers particularly if proceeding an unaspirated stop or anything more sonorant than a fricative. In the Deep Cave dialect these are always pronounced in their weak form in a syllable-final position.
/p, t, k/ > /β, z, j/ between vowels (/mäk/ + /ä:t/ > /mäjä:t) or voice to /b,d,g/ before nasals and glides (/mäk/+/men/ > /mägmen/) (/tut/ + /wär/ > /tudwär/). These also weaken and cause voicing or de-aspiration before stops (/mäk/+/tʰir/ > /mäjtir/) (/mäk/ + /tär/ > /mäjdär/). In the Deep Cave dialect these are always pronounced in their weak form in a syllable-final position.
/ t͡s/ > /t∫/ after an aspirated consonant and /ʤ/ after unaspirated consonants while the consonants themselves weaken as usual.
(this isn't the chart because I am *really* not confident in this choice)
/ɸ/ disappears after nasals and liquids but causes them to de-voice (/län/ + /ɸät/ > /län̥ät/ and /mur/ + /ɸät/ > /mur̥ät/)
/r/ is labialized for some dialects and becomes (/w/) before consonants (/därʷgirʷ/ > /däwgirʷ/) (/murʷ/ + /ɸät/ > /muwät/)
/n/ assimilates to the point of articulation of the consonant it precedes (/kän/ + /ɸät/ > /kämɸät/) (/kän/ + /ker/ > /käŋker/)
Word-final clusters:
There are no word-final clusters in roots. Clusters can occur following the same phonological rules as above via suffixes* /(kʰäpʰ/ + /tʰ/ > /kʰäftʰ/)
Word-initial consonant clusters: None.
(Was considering /gn/, /dn/, /bn/ but couldn't think of any way to justify it since they're kinda weird. My initial thoughts were along the lines having a vowel between unaspirated stops and /n/ disappear in certain compounds (/kun/+ /ɸät/ > /gnät/) and then reassert itself in further compounds (/mekʰ/ + /gnät/ > /mexkun̥ät/) but that seemed unlikely for probably many reasons not least being the stress on the first syllable.)
Vowels are strictly /ä, ä:, e̞, i, u, u:/
(Was leaning towards changing /ä/ to /æ/ but that would make the vowels really lopsided.)
The only diphthong is: /au/
Syllable structure of roots: CVC, VC
Affixes can be: -C, -VC, -CVC
Stress is always on the first syllable.
*These "single consonant" suffixes exist due to the loss of /h/ and then the loss of word-final vowels.
---
Some notes that might explain some of sounds above: The language initially only had the vowels /ä, e, u/ and also had /h/ which could appear at the beginning or end of a syllable. In places where the /h/ was lost and two similar vowels jutted together two-syllable words became single syllable words with a long vowel (käh.här>käär>kä:r). For a time the language also had /e:/ due to this same process but eventually /e:/ became /i/. Diphthongs created in this process eventually simplified or underwent other phonological processes and disappeared with the exception of /au/.
The /j/ that exists as an allophone of /k/ and the /j/ at the beginning of words are historically distinct. Word-initial /j/ is also part of the phonological changes that occurred after the loss of /h/ and the transition of /e:/ into /i/ (/heh.hät/>/ʔe:.ät/>/ʔi.ät/>/jät/) (/heh.hek/>/ʔe:k/>/ʔik/>/jik/). This is also where /w/ originated (/huh.häk/>/ʔu.äk/>/wäk/) as well as aspirated word-final consonants (/tuk.heh/>/tukʰ.ee/>/tukʰ/).
Word-initial /h/ became /ʔ/ which still persists in words that would otherwise be considered to start with a vowel (/häh.hät/>/ʔä:t/).
(So I may need to adjust my syllable structure to strictly CVC or change my chart... wasn't really sure how to count that.)
---
A few questions (?) or areas of uncertainty (?) and general notes:
1. I don't know how unusual it is to have a stop series with a 3-way distinction between ejectives, aspirated stop and unaspirated stops *and* allophonic voicing.
2. As mentioned above, I am really not confident with this attempt to resolve complex clusters involving /t͡s/... not that /xt∫/ is less complex really but I do find it easier to pronounce than /kʰt͡s/. Still, this is probably where I am the least certain of my choices.
3. I'd also really love to have /ʍ/ in here somewhere, not sure if phonological processes can stack up so that in dialects where /rʷ/ becomes /w/ before consonants the devoicing power of /ɸ/ still applies so you actually have /murʷ/ + /ɸät/ > /muʍät/ instead of /muwät/. If this is so then I can probably also add a voiceless /m/ in there which would be great (/kän/ + /ɸät/ > /käm̥ät/).
4. My word-inital clusters didn't work out but I am debating reversing it so I have /mb/, /nd/, /ŋg/ that resulted from /kun/ + /hät/ > /kund/ and then /kund/ + /wär/ > /kunädwär/).
5. As hinted above, the conculture that speaks this language live in caves so I wanted to have a lot of sounds that seems like they would suit the acoustics-bonus from that hence the ejectives and voiceless sonorants.
Sooo, all that being said- did I miss anything? Does it make any sense at all? Anything that needs further explaining? I tried to cover the things that stood out most to me but I also know I am leaning *heavily* on that disappeared /h/.
