the stuff between the commas

LFG List dalrympl at parc.xerox.com
Mon Aug 4 23:29:35 UTC 1997


OK, here's a go:

 > 1.  The airplane, I believe, arrived.
 > 2.  By God, and God is almighty, you bought the car!
 > 3.  We, the undersigned, protest.
 > 4.  At the show are cars, if you are interested, imports.
 > 5.  The friend is Peter, may God bless him.

The large LFG grammar of English that I wrote at Xerox some years ago now
handled precisely one of those cases, number 3, via a rule roughly
like this:

NP -> N'' 
      [ COMMA
        NP: v in (^ ADJ) (v CASE) ~= GEN
        (COMMA) ]*

I.e., the second appositional phrase was made an adjunct of the noun
phrase.  This seems to me not unreasonable, since it functions much like
a non-restrictive relative.

The same technique would extend straightforwardly to cases 2 and 5
providing one allowed as well as a noun phrase above various other kinds
of exclamative sentential constituents, perhaps as a CPexcl category
that expands appropriately.

Cases 1 and 4 are fundamentally different, because they are not attached
to an NP, but are rather free-floating elements withing sentences
("parentheticals").  For dealing with these, the fundamental problem is
what you think the f-/semantic structure should be.  If you think 1
should be equivalent to "I believe the airplane has arrived" then you
have problems -- because one is getting a kind of inversion between
semantic prominence and syntactic prominence that has been discussed
some in the literature (using terms such as "head-switching").  If
however, you see the role of "I believe" as like "probably", then this
can also be a sentential adjunct.  The only trick is to supply some sort
of pro object to believe to satisfy subcategorization

S -> ...
     ( COMMA
       S: v in (^ ADJ) (v {COMP|XCOMP}* OBJ PRED) = 'pro'
       COMMA )
     ...

Chris Manning.





More information about the LFG mailing list