Re: Allophone Problem

John Vertical wrote:

> Joseph Fatula wrote:
> (...)
> > Let's try again:
> >
> >- [tes] > [teza]
> >- [tiz] > [tiza]
> >- [kef] > [keva]
> >- [kiv] > [kiva]
>
> Well, you might explain that the lowering of /i/ is not triggered by the
> unvoiced fricativ *phones*, but the *phonemes*; ie. that [teza] is still
> underlyingly /tisa/,
Yes, and [tes], [kef] are /tis/, /kif/-- min.prs. underlyingly but not at
the surface. (I _knew_ it was something like this-- ya beat me to it...)

Would it affect /u/ as well? [tos] vs. [tuz], [toza] vs. [tuza]? If so a
very simple (though peculiar) phonetic rule:

(Early rule) Hi V --> lowered to mid / __[vl.fricative](V)
  produces tes < tis AND tesa < tisa (/tiz/ is not affected)

(Later rule) [-voi fric.] --> [+voi fric] / V__V
  produces teza < tesa (tiz-a is not affected)

Less peculiar, and more consistent IMHO, you might want the change to take
place before _all_ vl.sounds (assuming your structure allows them)-- just
restricting it to vl.frics. is odd. So you'd also have [tep] Vs. [tib],
[teba] vs. [tiba], [tek] [tig] etc.

The origin could well be something like the Engl. vd.cons/length rule--
peace [pis] // peas [pi:z]-- then the short V lowers, and the long V
shortens.

> I'd find it less contrived, however, to just
> consider them different phonemes with a quirky distribution.
>
Quirky but entirely predictable, therefore, not phonemic.