Re: Phoneme Question

On Sat, 4 Aug 2007 19:28:09 -0700, Joseph Fatula wrote:

>   Here we'll talk about a language that has 32 possible syllables,
>phonetically.  There are basically 16 phonetic realizations of
>consonants, and 8 phonetic vowels.
>
>   I could analyze this as a language with only 2 vowels that have
>allophones based on the preceding consonant, in which case there are 8
>consonants
(I think you mean "16 consonants")

>or I could describe this as a language with 8 vowels and
>only 4 consonants, which have allophones based on the following vowel.
>Which is it?  How can I tell?
>
>   The syllables are as follows, orthographical first, then XSAMPA:
>
>     "pa fa ma va - te se ne re - či ši ñi li - qu xu �?u wo"
>     "pâ fâ mâ vâ - tê sê nê rê - tî šî nî rî - kû hû �?û wû"
>
>     [pa fa ma Ba - te se ne 4e - cCi Si Ji Li - qu xu Nu wo]
>     [p6 f6 m6 B6 - t@ s@ n@ 4@ - tI SI nI 4I - kU hU NU wU]
Yeah, the thing is that you *can't* tell by this alone; too much
complementary distribution. You'd need to test whether nativ speakers
interpreted, say, [k@] as /t@/or /p6/ or /kU/. However, the triplets /cCI tI
t@/, /Ji nI n@/ & /Li 4I 4@/ suggest that at least the 16-consonant analysis
does not work. I would expect /cCI JI LI/ in that case... Also, [qu] strikes
me as a bit surprizing, what with [u] being closer to [k]. If you want more
distinction from [kU], why not [kw)u]? Or [?U] for the latter, for symmetry
with [xu hU]?

I'd *prefer*, OTOH, an analysis with 4 vowels and 8 consonants - merging the
velars + labials as a "gravis" and the alveolars + palatals as an "acute"
series; and merging /a 6 u U/ with /e @ i I/ resp. to have just ±tense & ±high.

Actually, maybe gravis/acute could be better considered a suprasegmental
feature here; you could whittle the numbers down to 4*4 then.

John Vertical

Reply