On Fri, 23 Apr 2004
enlil@... wrote:
>
> Phonetics and phonemics are different things.
Wow.
> [...] Using IE as an example, we write *sedtos but we know that there is
> evidence to show that two dental stops side by side caused an
> interloping s-sound. Thus *sedtos may be written phonetically as
> [setstos] [...]. Of course, that doesn't mean that we write *setstos,
> because the phonemes involved are really *d and *t and we know that in
> this rare instance where they come together, it is pronunced thus.
No, if the pronuncation really was that way, the phonemes involved are
really /setstos/. It would only be an even more abstract analysis that
allowed us to guess that it was "really" made up of elements justifying us
to write it //sed-to-s//. That abstract analysis, however, can be carried
much further, so that there is really no final abstract analysis. But with
the bi-unique convention there *is* a phonemic level whose form ought to
be beyond dispute. And that would stipulate /setstos/ here, provided, that
is, a word "really" made up of /sets-/ + /-to-s/ is pronounced the same.
If that is not the case, and /t/ does not fully merge with /ts/ when
followed by a dental stop, the correct phonemic analysis would be
/settos/.
There is another Danish example of multiple degrees of length, rhyning
with the one first given:
I. <spar es> 'ace of spades' is /sba Es/ with deletion of the glottal stop
of /sba?/ <spar> in the close juncture with a following accented word.
Since phonemic analysis is undertaken without consideration of the content
plane, this just establishes a word /sba/ for this language.
Now, besides /sba/ we have:
II. /sba:/ <spare> '(to) save, spare'
III. /sba::/ <sparer> prs. 'save(s), spare(s)', '(a) saver' (one who saves
money on a savings account).
IV. /sba:::/ <sparere> pl. 'savers' (who save money on a savings account).
All of these four words are fully lexicalized and generally have no other
form. None of them is abnormal in any way.
As before I am hesitant to accept a difference between the second and
third degree, but at least <spar> : <spare(r)> : <sparere> are certainly
different degrees of pure length.
Jens