[toolstick] True enough, but what does one learn about the basics of the 805x core from using that device?
EVERYTHING, the on chip debugging allow you to see every bit in the memory, SFRs, registers, ......

When starting with a new MCU, that "new ground" has to be plowed up anyway, and I find ASM inherently easier, since it is bound to the core.
only somewhat true. There is no difference in SFR handling between C and asm. It is inherintly more important to understand the SCON than the CJNE.

Do the SiLabs parts all have a completely standard 805x core,
yes

including UART and parallel ports on power-up?
derivatives do, deviates mostly.
They certainly require some additional setup if you want to use any of their "features" not that you'd have to do that.
NOBODY in his right mind would start a beginner on the PCA whether his chip has it or not

I'm still persuaded that using a 1985 or earlier i8752 would be very easy, though those antiques are quite costly. There'd be no misgivings about whether the features are standard, though.
WGASA, if you read the datasheet the features, whether antique or modern are well described. If you insist that a beginner start with a steam driven derivagtive, then your "learning curve" arguments just get delayed, not removed.

Hopefully the O/P has taken to reading lots of spec's, and maybe attempting to understand the instruction set and SFR/hardware interactions with it.
WGASA about the instruction set, the SFRs are the issue

The old-standard 5-volt 40-pin DIL-packaged chips are still the easiest to use when learning, methinks.
"learning" is a whole lot better accomplished starting out with a devboard, that allow you to learn the soft without worring about the hard till later (shortening the learning corve)

Erik