[developers] Transfer rule question

[Berthold Crysmann]

I think you need +copy+ for that.

B


On 04/03/2016 00:57, Emily M. Bender wrote:
> Dear developers,
>
> I'm wrapping up this year's iteration of Ling 567 and have come to a 
> question
> about transfer rules. One of our sentences is "Cats chase dogs and sleep".
> One of our languages is a VSO language, which doesn't have VP 
> coordination.
> It does have subject drop, so the authors of this grammar have 
> translated that
> sentence as "Tigers (!) chase dogs and [something] sleeps".  We wanted to
> see if we could write a transfer rule that would take the VP 
> coordination MRS
> from English and make the corresponding S coordination (with subject 
> drop),
> worrying only about the case where the second conjunct is an 
> intransitive verb.
>
> We got as far as:
>
> vp-coord-to-s-coord := monotonic_omtr &
>  [ CONTEXT [ RELS <! [ PRED "_and_coord_rel",
>                        L-INDEX #e1,
>                R-INDEX #arg0 ],
>                      [ ARG0 #e1,
>                        ARG1 #arg1 ] !> ],
>    INPUT [ RELS <! [ PRED #pred,
>              LBL #lbl,
>              ARG0 #arg0,
>                      ARG1 #arg1 ] !> ],
>    OUTPUT [ RELS <! [ PRED #pred,
>                       LBL #lbl,
>                       ARG0 #arg0,
>                       ARG1 x ] !> ],
>    FLAGS.SUBSUME < #arg1 > ].
>
> ... which does apply and doesn't spin. However, the coref tag #pred 
> between INPUT
> and OUTPUT don't seem to be doing any work.  That is, the ARG0 is getting
> copied across (the identity between the _and_coord's R-INDEX and the 'new'
> EP's ARG0 is there) but the PRED value is getting lost.
>
> Does the LOGON transfer formalism allow for preserving PRED values?  
> If so,
> what is the syntax I should be using?
>
> Many thanks,
> Emily
>
>
>
>
> -- 
> Emily M. Bender
> Professor, Department of Linguistics
> Check out CLMS on facebook! http://www.facebook.com/uwclma