Ilog-kwa
(ìloN kwa): My first serious attempt at a constructed language, started July 2006. A work in (very) slow progress.
Phonology
Ilog-kwa uses 17 Roman letters to represent 17 (or 19?) phonemes.
/p t k/ <p t k>
/m n N/ <m n g>
/l j w/ <l y w>
/s S h/ <s x h>
/ts tS/ <ts tx> (I am not sure if these are phonemes; they only occur across syllable boundaries).
/a e i o u/ <a e i o u>
Allophones:
/T/ <t-coda followed by stop or nasal>
/x/ <k-coda followed by stop or nasal>
/K/ <l-coda followed by h>
There is no phonemic voicing or aspiration distinction.
From here on, I will write in the orthography.
Phonotactics
Syllables may take these structures: (C) (C) V (C).
Legal onsets: null, any consonant, any non-approximant consonant + 1 approximant, l + w or y.
Legal nuclei: any vowel.
Legal codas: Any consonant, with the following execptions: p cannot be a coda. y cannot be the coda in a syllable with an e or i nucelus. w cannot be the coda in a syllable with an o or u nucleus.
Sandhi and Allophony
t and k codas are spirantized when they are followed by a nasal or a stop, i.e. t becomes /T/ and k becomes /x/.
When l occurs next to h in either sequence, the resulting sound is /K/.
The silibants have a strong tendency to be voiced when they occur next to nasals. The stops also tend to be voiced in certain positions. Because voicing is non-phonemic, "wrong" voicing will not alter the meaning of any word or utterance.