... in the accent on certain features useful for small control applications. The real microcontroller has maybe less developed memory access (as in small control applications there are typically small amounts of data and the code is relatively small, too); but the peripherals are bound more closely to the processing core for faster access to them.

Not that a microprocessor could not cope with control applications, it's just not as suitable for it as a dedicated microcontroller. This is for example the reason for slight disappointment some of the seasoned '51 users experience when they expect a superfast ARM-based chips to perform superfast I/O operations (e.g. pin toggle).

Not that a microcontroller could not cope with general purpose processing, it's just cumbersome in doing so.

At the end of the day, the differences come to surface in the applications pushing some limit - e.g. consumption, chip area...

I do have problems with terminology if taken too strictly, too; but I see some point in these two terms. We can talk about percents of relevancy... :-)

JW