Re: Graphemes defined

John Cowan wrote:

>
> Doug Ewell scripsit:
>
> > Regardless of origin, I would consider anyone who actually pronounces
> > "vision" as /vIzj@.../ or "measure" as /mEzj@.../ to be speaking an
> > unusually stilted dialect of English.
>
> The point is not that anyone says [vIzj@...], but that a certain degree
> of theoretical neatness is achieved by treating the underlying form as
> /vIzj@.../ and adding a phonological rule that realizes /zj/ as [Z].
> Note carefully the distribution of slashes and brackets here.
>
> These things change over time, to be sure; [S] once appeared only as
> one realization for /sk/ or /tj/, but over the course of time minimal
> pairs like "shirt" and "skirt", "dish" and "disk" (often, as in these
> cases, of the same ultimate origin) have arisen, and the phonemic
> status of /S/ is now perfectly clear. The same is not true of /Z/.

How is it "neater" to treat Z as /zy/ but not S as /sy/?

Next you're going to be claiming that C and J should be /tS/ and /dZ/,
or would you prefer /tsy/ and /dzy/. Phonemes are set up within
phonological _systems_, and the _system_ is much "neater" with /Z/.

> A few other such possibilities (without falling over the cliff into
> extreme Chomsky/Halleism): assuming that "strong" and "long" have
> underlying forms ending in /Ng/, as shown by "stronger" and "longer",
> even though realized as [N] finally; treating the range of Chinese
> vowels as underlyingly only /a/, /@/, and zero, with possible on-glides
> and off-glides of /j/, /w/, /H/, and zero.

"Underlying" has no place in this sort of phonemic analysis. It _is_
Halleism. (Chomsky bears little responsibility for "generative"
phonology.)
--
Peter T. Daniels grammatim@...