se jerna & seljerna, revisited

[Mr Andrew Rosta <ucleaar@UCL.AC.UK>]

Some weeks ago on this list I kept on harping on the following
two points;
  (1) Must _se jerna_ be synonymous with _seljerna_?
  (2) If not, one might wish for a way of deriving a lujvo
      from simply _jerna_ (as opposed to _se/te/xe jerna_)
      such that the x1 of the lujvo corresponds to the x1
      of _jerna_.

The upshot was that the answer to (1), coming from Lojbab &
Colin, was "no", though no convincing supporting arguments
were given (i.e. the No was basically stipulated), and that
the answer to (2) (from Lojbab) was that this desideratum
seemed rather pointless & that the closest thing was a
cmavo that signals loose usage.

I think I can now answer both questions satisfactorily.

(1) No. The proof is that while _se jerna_ must have
the place structure of _jerna_, with x1 & x2 swapped,
_seljerna_ can have an utterly different place
structure.

(2) The solution is _selseljerna_. This is exactly
what I was looking for: one would expect this to
have the place structure of _se se jerna_ = _jerna_,
but in principle the structure can be completely
different. This technique would work for all but
1-place gismu(s) - in fact, I imagine that it might
even work for 1-place gismu(s) as well.

----
And.