Andanic languages

Phonology

Old Andanese: /p m t n l k g h q kʷ gʷ hʷ ŋʷ qʷ/ for consonants, /a e i o u/ on two tones for vowels. Note that Old Andanese preserved the lack of /s/ passed down from the parent language. Note that all the fricatives (/h hʷ g gʷ/) are laryngeals. (The letter "g" always indicates a fricative; ġ is used for the stop in related languages but does not occur in any stage of Andanese.)

Allophony and sandhi

Old Andanese /p/ and /t/ became voiced between vowels.

In general, Old Andanese is "vowel-strong", meaning that vowels affect the pronunciation of consonants more than the other way around. This is in contrast to most of the languages of the Gold family, where consonants differ very little in pronunciation from one environment to another but have strong effects on the vowels that surround them. The Gold family's languages also tend to have more complex syllables than Old Andanese.

Representation of dorsal consonants

The language is very guttural compared to its neighbors; after all, it has 9 dorsal conosnants, but only 3 coronals and 2 labials (though five of the dorsals are labialized). However, the syllable strucutre is almost entirely CV, so it does not sound quite as aggressive as one might think.

Note, also, that /l/ patterns grammatically as a dorsal consonant, which means that from the viewpoint of the Old Andanese speakers, there were only four non-dorsal consonants in the language: /p m t n/.

Daughter languages

Old Andanese had a few daughter languages that survived, mostly spoken in Nama, before being overwhelmed by Khulls. THose that survived the longest tended to be found in the most isolated areas, but note that due to Nama's unusual geography, some of the most isolated areas were right along the coast.

Galai

Properly spelled G̣alai, as this language developed a true voiced stop /ġ/ alongside the fricative /g/ it inherited from Old Andanese. Galai is very conservative. /q/ > /ʔ/ > /0/ unconditionally; labialization was mostly dropped; intervocalic voicing became phonemic (because the consonants were restored to voiceless if stressed); the voiced fricatives /g gʷ/ hardened in initial position into stops.

• Because the language is vowel-strong, it may do /ti/ > /si/ (likely t > č > š > s). Possibly also /ki/ > /si/, and then /kʷi/ > /ki/.

• /q/ > /0/ may be ignored in monosyllabic words, or perhaps in initial position generally, in favor of shifting to /k/. This would prevent words consisting of just single vowels from arising. If so, this /q/ > /k/ must be made later than the /k/ > /s/ shift.

• Also, the syllable /qi/ might delete its /i/ early on, leading to a setup similar to that of Japanese, where geminate conosnants appear in place of the earlier /qi/. Possibly /qu/ also. Note, however, that this would not likely happen if there was not also a nasal coda other than that left by the remnants of the very rare syllabic consonants.

• Initial /d/ is rare, coming only from the sequence /ġi/ plus a vowel. There are probably no phonemic palatals in the language; if there are, they are like Japanese in that they cannot occur before /i/.

PROBABLE FINAL PHONOLOGY

/p b m/

/t d n s l/

/k ġ ŋ h g/

/a e i o u/

/ă à ā á/

/j/ and /w/ could be considered phonemes as well, despite being written as vowels, since they cannot carry tones.

Again, note that /g/ is a fricative and /ġ/ is a stop. /g/ is not likely to delete, despite being the only voiced fricative in the language. More likely is that it will simply merge back in with /ġ/, becoming allophonic (again, similar to Japanese).

A rare /r/ may come from the sequence /lj/.

Olati

Possibly an entire family, since these speakers are more isolated from their neighbors than those of the other branches. However, as the evolution time is short, the languages will be very similar to each other and possibly best considered dialects with somewhat divergent phonologies.

Olati is "softer" than the others in that it pushes consonants forward in the mouth, like Pabappa. Labialization of dorsals and palatalization of dorsals both occur. That is, /kʷ gʷ hʷ ŋʷ qʷ/ > /p v f m p/, but /k g ŋ/ > /s j n/. Both shifts are unconditional. /h/ stays /h/, however, and /q/ shifts to /k/. There are almost no changes of vowels or tones. None of the consonants shifts to hiatus, at least during the stage of common Olati before the dialects broke up.

Lyugi

A language that changed more than the others. Spoken in the mountains of Repilia. Possibly a "Tarise" language.

• High vowels were deleted in unstressed syllables in some environments, creating consonant clusters. They were never deleted after labialized consonants, however, or when long (though all long vowels were stressed anyway). There are now palatalized and labialized versions of consonants, and these can occur before other consonants, but the plain (/a/) consonants do so only very rarely. (e.g. /top/ "blood" was native.)

• The labialization is deleted in most such clusters, while the palatalization moves towards /s/-like sounds and then is also deleted. However, before a vowel, labialization survives and then changes to a true /w/. Also, /q/ resisted palatalization.

• Word-final consonants in unstressed syllables are dleeted. (These words had two unstressed syllables in a row.)

• Word-final consonants in stressed syllables are often deleted, changing the tone and length of the vowel.

• Other clusters are generally resolved either by creating geminates, wholly new consonants, or making the preceding vowel long.

• Perhaps /e o/ > /i u/ in environments where the previous /i u/ had disappeared. (c.f. some dialects of Greek) And perhaps in fact it is unconditional, such that unstressed syllables can only have /a i u/. Again this would not affect long vowels because they are always stressed.

• /ʷa ʷe ʷi ʷo ʷu/ > /o o u u u/ (?) Some of these might be long vowels.

• /e o/ > /ə/. Thus, this is now a four-vowel language.

• /k/ > /c/, perhaps going further to /č/, /š/, or even /s/, when before /i/.

• /q/ > /k/ unconditionally. (Thus, it does not delete itself.)

• Intervocalically, /g/ > /0/ and /b/ > /w/. (/d/ > /r/, but this is considered just an allophone.) This may have been influenced by Khulls' shift.

• The resulting vowel sequences change to single (usually long) vowels. The changes are again similar to those of Khulls, but definitely unrelated, as Khulls had done its shift more than a thousand years beforehand. Note that /o/ now only occurs where there was once a /b/, and /e/ only occurs where there was once a /g/. Possibly /ja/ > /e/, but /ai/ > /ē/ (also as in Khulls).

• There should be more sound changes pushing towards /s/. All of the palatals, for example, up above could be simply merged together as /s/. (Escept nasals.)

PROBABLE FINAL PHONOLOGY

/p m/ <--- Note the lack of /b/ (because it was always intervocalic(?)) and of /w/ (considered just an allophone of /u/).

/t n s l r/ <--- Likewise /d~r/ is considered one phoneme. /s/ arises from multiple paths.

/c/ <-- Possibly just merges to /s/.

/k g ŋ h/ <--- Note that /g/ survived in initial position, and did not merge to any other sound.

/a e i o u ə/ <-- tentative. Schwa might merge to /o/ as in Japanese, to /u/ as in Khulls, or to /a/ as in Moonshine.

NOTE, also, that /p/ and /t/ will both be very rare phonemes in this language, and that might be a problem.