Talk by Matt Watson
A Critique of a Proof-Theoretic Treatment of Anaphora
In this paper, I will critically examine one recent proposal (Ranta
1991) to treat, within a proof-theoretic framework, the semantics of
anaphora in English. I will argue that this theory wrongly predicts, for
example, that the pronoun 'it' and the noun phrase 'a pen' can be
anaphorically linked in the following sentence:
- If every man finds a pen some man loses it.
Both discourse representation theory (DRT) and dynamic predicate logic
predict that this link cannot be made. Since the new proof-theoretic
approach disagrees with these 'venerable' model-theretic approaches in the
analysis of sentences such as this, much of my argument will center on
them. But I will also look at sentences involving anaphora where the
proof-theoretic approach and the approach of 'classical' DRT agree, but
where both approaches get the wrong results. I will suggest ways in which
the proof-theoretic approach can be modified to deal with these sentences,
following some suggestions in the DRT literature. Finally, I will give an
assessment of the overall prospects for this approach.
Paul Dekker, November 2, 1995