John Cowan wrote:
> Roger Mills scripsit:
>
> > Which is exactly what written English does. I've suggested in that past
> > that the _underlying_ phonology of _all_ Engl. dialects (the
standard-ish
> > one, at any rate) does have /r/ in all the positions where it is
written.
>
> I think you are right, but there are difficulties. What about ARSE, for
> example? The universal North American pronunciation is [&:s]; does this
have
> underlying /r/ or not? Surely you don't want to claim that BASS, the
fish,
> is underlyingly /bars/ or /b&rs/ (OE "baers").
In terms of the written lang. we'd just have to acknowledge that a few words
vary, even if their pronunciation obeys some other regular rule-- [&s] vs.
[A:s] (or whatever) is in the same class as [gr&s]-[grA:s] or [b&T]-[bAT]
etc. Is RP "bass (fish)" [bAs]? I seem to recall that not all US [&] are
[A] in RP, some are still [&]
>
> Final /r/ mostly patterns like pre-consonantal /r/, but what are we to
> do with "intrusive r" in non-rhotic dialects, where unhistorical r's
> appear in external sandhi by analogy to the historically lost ones?
>
It is simply not allowed!! :-( (Seriously, I don't know)