iron: /@r/ vs syllabic /r/

[Laurence Horn]

At 12:23 PM +0100 8/14/00, Aaron E. Drews wrote:
>>RP is underlyingly whatever it is, but it is surfacely non-rhotic
>>except in environments
>>that yield a linking rhotic realization.  Larry and I say "Bring back SPE!!!"
>>DMLance
>
>I suppose you mean 'linking' as in phonological sandhi.
>Traditionally, there has been a distinction between 'linking-r' and
>'intrusive-r', the linking phenomenon being in historic/orthographic
>instances of (r), and intrusive-r being the Cuber is and drawring.
>
>And RP isn't consistently 'rhotic' in a linking realisation.  It
>fluctuates because of social stigma.

Yes, I meant (and I assume DG meant) linking-r, which was why I
mentioned such alternations as "the car" vs. "the car is..." rather
than "Cuba" vs. "Cuber is".  It's only if at least some vowel-final
entries DON'T alternate in this way that you could really have an
argument for an SPE-style abstract segment that deletes in the
absence of vowel-initial suffixes and following words.  Actually,
it's really a Sapir-style argument, given his account of the abstract
/t/ phoneme in that example of /dinit/ (where it shows up in
alternations).

>
>As for SPE, it has a lot going for it, but I don't like the idea that
>I'm carrying around pre Great Vowel Shift phonology in my head.  I
>have enough problems coping with three (major) varieties of English
>on a daily basis.
>
Yeah, they do get carried away a bit.  The late Ted Lightner was the
best (i.e. most outrageous) exponent of the reductio ad absurdum
style of SPE analysis; he related "night" and "nocturnal", and
probably "hemp" and "cannabis", by synchronic rules of English
phonology.  I always had a weakness for the SPE analysis of "right"
as /rixt/, not to mention "boy" with an underlying mid-front rounded
vowel.  But you can buy "car" with a final -r in
non-rhotic/alternating dialects, or e.g. Fr. "heureux" with
underlying -z, without shelling out for these more extreme cases.

Larry