[b-hebrew] Generation grammar and b-hebrew

James Christian jc.bhebrew at googlemail.com
Sat Jun 19 16:43:09 EDT 2010


>   Syntactically all three are well-formed: NP
> subject, VP predicate with NP direct object followed by a string of
> properly formed adjuncts
> (in this case, PP's).
this kind of almost touches on the motivation behind the LFG formalism. In
LFG, in addition to the c-structure there is an f-structure. The f-structure
encodes functional information about the constituents. For example, as you
noted in English the noun phrases of a sentence are the same in structure
but functionally some are subjects and some are direct objects. The
f-structure in the LFG formalism encodes this kind of information and
proponents of the formalism feel that such descriptions behave better
cross-linguistically than simple c-structures in isolation.

However, I thought you might find this an interesting line because of the
strict separation between c-structure and f-structure information. To see an
example of what these structures look like check out this link:


Select English in the drop box and type a simple sentence like "Mary hit the
dog" and you get presented with a CFG tree (the c-structure) and an
attribute value matrix (the f-structure) which contains functional
information like which argument of the predicate is the subject and which is
the object.

Obviously, there is no direct way of doing this in Prolog. The technology
behind this web interface is the Xerox Linguistic Environment (XLE) which
isn't available freely as Prolog is.

James Christian

More information about the b-hebrew mailing list