I posted this over on the "random phonology/phonemic inventory" thread, but I want to apply it to Proto-Gadar, the ancestral language to the Gadar languages (spoken in 1AD in the orange region):
Consonants
/p t ʔ/ (P) <
p t q>
/b~β~m d~ɾ~n h₁~ŋ/ (B) <
b~m d~n g~g̃> (<
m n g̃> are used to represent the nasal allophones)
/s h₂/ (F) <
s h>
/l j/ (R) <
l j>
Consonants are divided into 4 groups based primarily on their manner of articulation, noted as P (voiceless plosives), B (voiced plosives ~ voiced fricatives ~ nasals), F (fricatives), R (approximants) as a shorthand.
/h₁/ and /h₂/ are phonetically identical when /h₁/ isn't a nasal, both being realised as [h]. Further explanation shall be covered below under allophony.
Vowels
/i/ <
i>
/(ə) o/ <
(e) o>
/ã/ <
a>
/ã/ is always nasalised, while nasalisation of other vowels is purely allophonic.
Syllable Structure
Syllables are strictly CV, with onsets being obligatory. No consonants can appear as a coda and consonant clusters are prohibited.
Stress
Stress is penultimate. In non-compound words, "secondary" stress appears every even syllable back through a word after primary stress, e.g. ˌCV.CVˌCV.CVˈCV.CV. Syllables carrying stress further back in the word act in the same way as final stressed syllables.
Stress is primarily realised as a mix of vowel length and pitch, with stressed vowels being slightly longer than unstressed one. Pitch rises through a word, peaking on the syllable carrying primary stress where the level drops down again.
Allophony
Palatalisation
/i/ causes palatalises preceding consonants to a certain degree, but this is most noticeable for the alveolar sounds /t d~ɾ~n s l/ which are realised as [tɕ dʑ~ʑ~ndʑ ɕ ʎ̟] respectively. This palatalisation is not indicated orthographically.
Nasalisation
/ã/ causes preceding "B" to become nasal which in turn causes the preceding vowel to become nasalised, e.g. /tibã/ > [ˈtɕĩmã]
tima.
/ã/ also causes following "B" to become nasal, e.g. /pãdi/ > [ˈpãndʑi]
pani.
Note, however, that nasalisation of the following vowel does not occur in this instance (not [ˈpãndʑĩ]). Similarly, "B" preceding nasalised non-/ã/ vowels do not become nasal, e.g. /doh₁ã/ > [ˈdõŋã]
dog̃a, not [ˈnõŋã]
nog̃a.
Spirantisation
Between vowels (when neither vowel is /ã/), "B" consonants become voiced fricatives, e.g. /ʔobi/ > [ˈʔoβi]
qobi
The Two Hs
The difference between /h₁/ and /h₂/ lies in the former being subject to nasalisation while the latter is not, with /h₁/ deriving from older
*g. /h₂/ also appears in order to break up vowel clusters.
/h₁/ thus appears as [ŋ] when adjacent to /ã/ while /h₂/ appears as [h]. There is no synchronic indication of whether an [h]~[ŋ] should be present in a word or not and must be learnt by speakers of the language.
The Schwa
/ə/ is an epenthetic vowel appearing to break consonant clusters. It is typically pronounced as [a] but without causing any nasalisation, either preceding or following, although it is subject to nasalisation. When stressed it appears instead as [*i] but does not trigger preceding palatalisation.
The result of this allophony is that it can produce minimal pairs such as [ˈti] vs. [ˈtɕi] or [ba] vs. [mã].
Orthographically this sound is always represented as <
e>, regardless of surface pronunciation, thus [ˈti] appears as
te and [ba] as
be
Further Vowel Allophony
/i/ and /o/ are subject to changes affected by both stress, position in the word, and the vowels of surrounding syllables.
/o/
- Stressed: [o]
- Unstressed
- - [*u ] when the following syllable contains /i/
- - [*u ] when word-final
- - [o] elsewhere
/i/
- Stressed: [*i ]
- Unstressed:
- - [ẽ] when the following syllable contains /ã/
- - [*i ] elsewhere
As mentioned above, whether stressed or unstressed, all instances of /ã/ cause the vowel of an immediately preceding vowel to become nasalised.