Stress and vowel reduction in Lindiga

Lindiga seems to have a mora-based accent pattern. The third mora from
the end is stressed, and the immediately preceding and following morae
are unstressed. An example in the modern language is the word "iakʉi"
meaning "at last, finally". The unstressed vowels are reduced in length
and essentially act like consonants.

i̯-ˈa-kʉ̯-i -> [ˈjaɡɥi]

According to the current phonology description (which hasn't been
updated on the web page for quite a while, so the web page is
inaccurate), the "k" in "iakʉi" should be voiceless. But if the "ʉ" is
acting as a voiced consonant, like "l" or "r", that would explain the
voicing. Another example is the word for "bitter", "kaliki".

ˈka-li̯-ki -> [ˈkaʎiɡi] ~ [ˈkaʎɡi]

The "official" explanation for the voiced "k" in "kaliki" is that a stop
is voiced if it follows an unstressed syllable. But the word "sʉləki"
("hideous") is a counterexample, with a voiceless "k" following an
unstressed syllable. (It also seems to be the case that "ə" only occurs
in unstressed syllables, so it could be an allophone of a different
vowel, probably /ɛ/.)

Exceptions to this pattern do exist.

muaɭi "shop, store" = [ˈmwaɽi], not *[ˈmuaɽi].
siəʂʂa "moon" = [ˈʃiəʂʂa], not *[ʃiˈəʂʂa].

Exceptions of the first kind might have originally contained a long
vowel or a consonant that was lost, but exceptions of the second kind
are not so easy to explain. It may be that whatever process created the
centering diphthongs (iə ʉə uə) took place after the stress position was
fixed. These kinds of clues might help me to reconstruct some of the
history of Lindiga.

Reply