Peter T. Daniels scripsit:

> > All three analyzable as /zj/.
>
> Absolutely not. This is H. A. Gleason, this is ultra-orthodox
> descriptive linguistics. Such "analysis" wouldn't even be invented until
> half a decade later.

I didn't say that *Gleason* could analyze them as /zj/.

> You would do well to read Gleason on phonemic analysis and minimal
> pairs!

Perhaps I would, but I do not see the particular significance of
monosyllables in minimal-pair analysis.

> > I will bet there are dialects in which all three of these show /dZ/.
>
> We aren't describing a "dialect." We're describing General American, in
> particular as instantiated by Henry Allan Gleason, Jr.

It is preposterous to claim that General American is not a dialect.

> > 1) Whether in some dialects all instances of /Z/ are analyzable as /zj/.
> >
> > 2) Whether there is any dialect in which /Z/ is *not* analyzable as /zj/
> > apart from limited lexical exceptions without minimal pairs.
>
> As you can see from the above remarks, such questions are entirely
> dependent on the linguistic theory you espouse. They are not questions
> of fact.

Whether the Earth orbits the Sun or the Sun orbits the Earth is also
theory-laden, but we are reasonably confident of the correct answer
nevertheless. The notion that "rouge" is underlyingly /ruzj/ is
absurd.

--
XQuery Blueberry DOM John Cowan
Entity parser dot-com jcowan@...
Abstract schemata http://www.reutershealth.com
XPointer errata http://www.ccil.org/~cowan
Infoset Unicode BOM --Richard Tobin