My recollection of the research/PWB split (I wasn't in either center) was that research (wisely) decided not to get into the business of software support or committing to a stable system, so PWB forked off to be a supported, stable version of UNIX. It was used by many of the computer centers within the Labs because of the promise of support and stability. Support included published manuals, with the hand of Ted Dollota (whose first language wasn't even English) creating some amazingly good documents. There was some idea exchange between research/and PWB (like the Mashey shell and the Bourne shell incorporating support for what is now thought of as "here documents"), but research was focused on innovation and PWB was focused on support. AT&T management no doubt were looking to an OS on their (IMHO mostly disastrous) entry into the computer business, for which they agreed to be broken up.

On Tue, May 15, 2018 at 3:48 PM, Lyndon Nerenberg <lyndon@orthanc.ca> wrote:
I think you mean 'style' and 'diction'. I thought those came from research? I
remember seeing papers about them in a manual; maybe 7th Ed or 4.2/4.3BSD?

IIRC they were written up in the BSTJ "Unix" number.

Similarly with learn: I have a vague memory of seeing it with BSD, but I
thought it came from 6th or 7th edition. A quick look shows a copy in 7th Ed.

I remmber learn being present on at least one early Xenix release I played with.  I've long forgotten the details of which release, though.