MJR> How about "see her" vs. "singer"?   (In my 'lect, at least, the 'i'
MJR> in "-ing" is [i], not [I]).

CG> Examples involving word breaks are always a bit weak for me ;))) .

Picky, picky. :)

MJR> I don't, however, accept that a native
MJR> speaker's intuition is sufficient to prove it.  Sometimes native
MJR> speakers recognize allophonic differences (even if they can't
MJR> describe exactly what the difference is), but that doesn't make
MJR> such allophonic pairs distinct phonemes.

CG> Example here. Because I've never met any speaker of any language
CG> recognising allophonic differences in his L1. When they recognised
CG> differences, it always ended up as phonemic differences.

Well, let me back up a bit; maybe I'm overstating.  How do you
determine whether two phones are allophonic variants of a single
phoneme, vs. distinct phonemes in complementary distribution?
Into which category do Japanese [h]/[p\] fall?  What about [s]/[S]
in the same language?