iron: /@r/ vs syllabic /r/

[Rudolph C Troike]

>From an old-fashioned (how times change!) SPE perspective, it is the most
likely solution to posit /r/ underlyingly and generate the [@]
phonetically (of course the rules would have to be ordered to create the
/ay/ first from underlying tense //i//). [I once proposed a "diaphonemic"
generative system to link regional varieties via a sort of shallow-level
comparative reconstruction which aproximated the SPE results; it's in the
Harold Allen readings.] So Aaron is right in this.
        From a more surfacy comparative point of view, it makes sense to
recognize the /@r/ as a sequence, since then r-less varieties can be
derived by dropping the /r/, but leaving the /@/. I'm guessing, without
checking, that at least some versions of RP in England might drop the
retroflexion, leaving /ay at n/. Does this ever converge with <ion>?

        Rudy