|
|
Greg Lee wrote:
> me wrote:
>> ...
>> For example:
>> (1) [tumbi] + [kaI] = [tumbik:ai]
>> (2) [il.an_] + [ko:] = [il.aNgo:]
>>
>> In the first word, [k] becomes [k:] (or [k] with delayed onset; both
>> pronunciations are acceptable). In the second word, the [k] is not
>> preserved; the [n_] becomes [N] and [k] becomes [g]. In what ways would
>> the
>> difference between (1) and (2) affect phonemic analysis?
>
> In no ways. These complex forms seem to display
> the same distributions of sounds as found in single
> morphemes, and the phonemic import of the distributions
> seems also the same. The gemination of the [k] is
> interesting, of course, but I don't see how it affects
> what one says about the phoneme system.
>
>> (In English, tea
>> + cup gives [ti:kVp] and if tin + can gave tincan, the pronunciation
>> would be [t_In_k&n_], not [t_INg&n]).
>
> So Tamil doesn't share with English this assimilation
> in position across word boundary. That's interesting,
> but English would have the same phoneme system
> even if it didn't permit this assimilation, I guess, so
> why do you bring it up?
Well, we have a choice between there being separate phonemes /k:/ and /g/
and there being a single phoneme /k/ and there are possible still other
choices when arriving at a phonemic inventory. In Tamil orthography,
there's only <k> and <kk>/<k k> but ignoring orthography, how would it be
determined whether there are three phonemes, /k:/, /k/ and /g/ or two
phonemes /k:/ and /g/ or one phoneme /k/? Considering both orthography and
pronunciation, how would it be decided whether two orthographic <k>s are
the same phoneme or different phonemes if they're pronounced differently in
the same context?
mun2 + kai would give mun2kai [mun_kaI] but
iLan2 + kO doesn't give iLan2kO; it gives [il.@Ngo:].
> Greg
|
|