It's probably been about 2 years now since I've posted anything in this thread. Anyway, I've been thinking about rebooting Project Steppe. I'd like to call it something else, but the original name of the language, Kwenti, sounds a little too Elvish to me now.
Here's something I've come up with for the new version of the language. I'm not sure if I like it.
/p t k kʷ/ <p t k kw>
/s h/ <s h>
/m n ŋ ŋʷ/ <m n g gw>
/j w/ <j w>
/i iː u uː/ <i ī u ū>
/ɛ ɛː ɔ ɔː/ <e ē o ō>
/ɑ ɑː/ <a ā>
/ɑ ɑː ˈɑ ˈɑː/ <a ā á ā́>
C = any consonant
V = any vowel
C = any consonant
The sequences /kʷu ŋʷu wu ji/ are all disallowed. Essentially any two consonants can form a cluster. However, /kw ŋw kʷw ŋʷw ts/ are resolved as [kʷ ŋʷ kʷ ŋʷ t͡s]. Vowel clusters are not allowed. Two identical vowels, ignoring length, become a long vowel. For example, /ii iːi iiː/ > [iː iː iː]. When two non-identical vowels come into contact, if the first is /i(ː) ɛ(ː)/, /j/ is inserted between them. If the first is /u(ː) ɔ(ː)/, /w/ is inserted between them. If the first vowel is /ɑ(ː)/, /l/ is inserted between them. For example, /ɛːu ɔi ɑuː/ > [ɛːju ɔwi ɑluː].
Stress is unpredictable. Underlyingly, all roots have a single stressed syllable. Some affixes attract stress, while others are unstressed. Stressed syllables tend to be pronounced with a higher pitch than surrounding syllables.
Before other consonants and word-finally, /kʷ ŋʷ/ are often realized as [kʷʊ̆ ŋʷʊ̆].
/ɛ(ː) ɔ(ː)/ approach [e(ː) o(ː)] word-initially, word-finally, and, for some speakers, in all open syllables.
/i(ː)j ɛ(ː)j u(ː)w ɔ(ː)w ɑ(ː)l/ often become [iː ɛː uː ɔː ɑː] word-finally and before consonants.
Before /j/ or syllables containing /j/, /kʷ ŋʷ w u ɔ ɑ/ are sometimes realized as [kʉ̯ ŋʉ̯ ʉ ɞ ä]. Similarly, before /w/ or syllables containing /w/, /j i ɛ/ are sometimes realized as [ɥ y œ].
Most consonants are lightly palatalized before /i j/.
Stops are usually unreleased word-finally and before other stops, oral or nasal, and the affricate.
Following nasals, and occasionally when between vowels, obstruents are often voiced. Vowels are lightly nasalized before nasal consonants.
Geminate obstruents tend to be somewhat aspirated, and geminate sonorants are sometimes devoiced.
/l/ is pronounced as a fricative by many speakers when geminated and when adjacent to obstruents. For some speakers, /l/ is more rhotic, but for most, it is something like [ɫ~lˤ], especially around /ɑ(ː)/.
The exact realization of /h/ can vary greatly from speaker to speaker, but it is almost universally post-velar.
<tótko> /ˈtɔtkɔ/ [tɔ˥t̚ko] "rope, cord, connection"
<kallúl> /kɑlˈlul/ [kɑɬˈɬˤul˥] "falcon"
<mī́gi> /ˈmijŋi/ [ˈmʲĩː˥ŋʲi] "moon"