[Prev][Next][Index][Thread]

RM stuff

> %TER: 	noun( n(printer,countnoun),
> %TER:               noun_modifiers( pre_modif([]),
> %TER:                               post_modif( <'by the desk'_struct> ))))

> =KEN: I disagree.  This is a trivial point, but the 'by the desk'
> =KEN: structure should appear within a np_postmodifiers() structure 
> =KEN: as the last argument of the entity() structure, not as a 
> =KEN: post_modif() within the noun_modifiers() structure.  (I know,
> =KEN: that's not the point here).

Ken's right.  I framed the movement in terms of existing structure and
that's just plain wrong--post_modif/1 is for 'elect' in 'president elect'.
So reattachment will need to insert structure de novo from a catalogue of
patterns.  The right construct according to current DIPETT grammar would be:

% .. np_postmodifiers([pp( Prep, Phrase, PTs ),
%                      parenthetical(conj_pps( Links, PPs ))])
% .. np_postmodifiers([pp( Prep, Phrase, PTs )])

where PTs are the PP Tokens. 

Perhaps since we are no longer in parsing with its objective of exactly 
describing syntactic structure, the set of acceptable representations
might be extended to:

     np_postmodifiers([pp( Prep, Phrase, PTs ), PP ... ]).

It is easy to envision 'the printer by the desk in the office at the end 
of the hall'.

**SYL: Ken is trivially right (;-^)! I didn't mean to parse Terry's
**SYL: suggestion and I should have done so, apparently. Gee, I tell
**SYL: you, those Ph.D. candidates today ...
**SYL: One remark concerning your last point Terry: I would be inclined
**SYL: to stay away from modifs that could throw in some confusion about
**SYL: the parse structures. At least in the beginning. Let's keep
**SYL: verification as simple as possible.

###############################################################################

> %TER: We all know that with the exception of positional NPs, PPs are the most 
> %TER: frequent markers of Cases. I wonder if we could then view the conj_pp
> %TER: list in a complement as a list of CMs once reattachment had properly
> %TER: subordinated its elements to one another.
> 
> *SYL: Once reattachment has been performed correctly, a verb's
> *SYL: complement conj_pp should correspond to a list of CMs (and fillers). 
> *SYL: However, a noun may well have its own conj_pp (as postmodifiers) and
> *SYL: these shouldn't be identified as Cases, obviously.
> 
> =KEN: True enough... but as Terry mentions, the list wouldn't include
> =KEN: positional markers (or adverbial markers either).  What would be
> =KEN: the use of such a list, especially when CA gives us a complete
> =KEN: CM list?

The remark wasn't meant strictly, just as an observation--here lies treasure.  
Haiku will need to be rewritten though to operate on decomp output or the 
benefits of reattachment will be lost.  This should be an easier task than
the first time around?

**SYL: My impression was that Terry was looking
**SYL: for another name maybe more meaningful to the user of the RM.
**SYL: If it's just an observation ...
Follow-Ups: