11.4.7 Pronoun Resolution

The DRT implementation is able to resolve anaphoric pronouns as well, including those appearing in donkey sentences.

The DRT implementation is able to resolve anaphoric pronouns as well, including those appearing in donkey sentences. Let's have a look at some examples (the second example is actually an instance of a donkey example):

> A man walks. He smokes.
 
 
1 drs([A],[smoke(A),male(A),walk(A),man(A)])
1 exists(A,smoke(A) & (male(A) & (walk(A) & man(A))))
 
 
> Every man that loves a woman likes her.
 
 
1 drs([],[drs([A,B],[love(B,A),woman(A),man(B)])>drs([],[like(B,A),female(A)])])
1 forall(A,forall(B,love(B,A) & (woman(A) & man(B))>like(B,A) & female(A)))

How did we implement this? Let's have a closer look at the semantic macro for pronouns:

proSem(Sym,[arg:[index:X],drs:Drs,ana:Ana]):-
   Drs = [in:[drs(D,C)|S],
          out:Out,
          restr:_,
          scope:[in:[drs(D,[Cond|C])|S],out:Out]],
   compose(Cond,Sym,[X]),
   Ana = [in:A,
          out:[accessible(X,[drs(D,C)|S])|A]].

At first glance this semantic macros bears a great resemblance with the semantic macro for proper names. But there are two differences. The first difference is that there is no new discourse referent introduced. The second difference is that we add a contraint accessible(X,...) to our list of binding constraints. This constraint requires that there must be an accessible discourse referent for Xin the incoming DRS.