[logon] reflexives

[Stephan Oepen]

hi francis (copying the `logon' list, in case others are interested),

> I am running what I think is an up to date logon CVS, which I am
> expecting to change into SVN at any moment.

yes, as of this morning, CVS is history :-).  the CVS repository is now
write-locked, and i am in the process of preparing the SVN universe.  i
plan to announce SVN access early next week, for us to test the HandOn
release.  once that is complete, i can open the SVN trunk for component
maintainers to commit new revisions.

> I tried to use a new rule template: arg1refl_arg12pron_mtr, and I
> can't get it to work?
> 
> jisatsu-kill+oneself := arg1refl_arg12pron_mtr &
> [ INPUT.RELS  < [ PRED "ja:_jisatsu_s_1_rel" ] >,
>   OUTPUT.RELS < [ PRED "_kill_v_1_rel" ] > ].
> 
> arg1refl_arg12pron_mtr := monotonic_mtr &
> [ INPUT.RELS < [ LBL #h0, ARG0 #e1, ARG1 #x2 ] >,
>   OUTPUT [ RELS < [ LBL #h0, ARG0 #e1, ARG1 #x2, ARG2 #x4 ],
>                   [ PRED pron_rel, LBL #h5, ARG0 #x4 ],
>                   [ PRED pronoun_q_rel, ARG0 #x4, RSTR #h6 ],
>                   [ PRED identity_rel, LBL #h5, ARG0 #x2, ARG1 #x4 ] >,
>            HCONS < qeq & [ HARG #h6, LARG #h5 ] > ] ].
> 
> My understanding is it takes an input which is just an intransitive
> verb (ARG0, ARG1) and creates X verbs X-self.
> 
> When I try to use it, the output is pink, and I can't generate.
> 
> As far as I can tell, the new ERG doesn't actually have an identity_rel.
> 
> Won't this break things for noen?

those `identity' EPs are a convention we invented in LOGON to allow the
norwegian grammar to record co-referentiality (which is grammaticalized
in norwegian), without actually equating the `pron' varibles.  this was 
mostly owed to the realization that breaking up variables (for the ERG,
where co-referentiality is not encoded in MRSs given to the generator)
is more difficult than trowing away an extra EP.

at present, the downstream mechanism that maps `identity' EPs into the
ERG [ PRONTYPE refl ] is in `uio/noen/erg.mtr', i.e. not part of what i
have started to call the Transfer Matrix (types and accomodation rules
shared across language pairs).  i believe i should move it up, so that
your example above should then work (i expect).  you could give that a
try locally in the meantime ...

--- but first i need to switch over to SVN myself.  no more commits to
that ancient CVS tree at this point!

                                                     all best  -  oe

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++ Universitetet i Oslo (IFI); Boks 1080 Blindern; 0316 Oslo; (+47) 2284 0125
+++     CSLI Stanford; Ventura Hall; Stanford, CA 94305; (+1 650) 723 0515
+++       --- oe at ifi.uio.no; oe at csli.stanford.edu; stephan at oepen.net ---
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++