Re: Rotokas (was: California Cheeseburger)

Nik Taylor said:
> "Mark P. Line" wrote:
>> If you have a phonological analysis of Rotokas that shows how it's
>> better
>> to have separate phonemes for long vowels, I'm all ears. The only
>> analysis
>> I've seen posits 11 phonemes -- 6 consonants and 5 vowels.
>
> I don't know much about Rotokas.  I only know that it has been analyzed
> with far more syllables, and so have concluded that there must be some
> reason.
I don't think we have any evidence that Rotokas has ever been analyzed
with so many syllables. Seeing the number '350' in a single *secondary*
source is not a good data point when we can find in a primary source that
Rotokas has 11 phonemes and that its syllable structure is only (C)V. Even
counting the extra consonant grapheme mentioned earlier, it's impossible
to come up with anything remotely approaching 350 syllables. It's far more
likely that the correct computation was made for (C)V (which would be 35
syllables in this case) and simply misprinted (or miscopied by somebody
who intuitively disbelieved the correct value) as '350'.


> I seem to recall that that page mentioned that long vowels were
> phonemic.
It says there are 5 vowels in Rotokas, each of which occurs phonemically
lengthened.

There are a number of ways to interpret that claim. One way it *can't* be
interpreted is that it posits 5 more vowel phonemes, because the same
paper is very clear about the number of phonemes (5V, 6C) in the language,
with minimal pairs provided.

The most likely interpretation here of vowels being phonemically
lengthened is that the vowels are not non-phonemically lengthened (as they
are in English). That simply means that a phonetically long vowel
contrasts phonologically with a phonetically short vowel. It doesn't
entail that phonetically long vowels realize a single phoneme. From the
phoneme inventory given, it's obvious that a phonetically long vowel like
[a:] in Rotokas realizes a pair of phonemes /aa/. (If it had been said
that Rotokas vowels occur *non-phonemically* lengthened, then [a:] would
just be an allophone of /a/. This is what we have in English.)


>> > I'd guess that [s] is an allophone of /t/ before /i/.
>>
>> It could also be an allophone of any other phoneme, if it's an
>> allophone.
>
> But /t/ (or /k/, come to think about it) is most probable, and many
> other languages in that family have [s] as an allophone of /t/ before
> [i]
I didn't know that was common in East Papuan. Do you have a reference?


>> The question would still remain as to why it's written as 's' and not as
>> 't'.
>
> Why not?  The standard romanization of many languages includes
> subphonemic distinctions.
I can't think of any examples right now in modern, phonologically
engineered orthographies. Can you remind me of some?


-- Mark

Replies