ABSTRACT
So far, I have concentrated only on the structural conditions governing anaphora. A major question still to be answered is what is actually meant by anaphora or coreference. While much attention has been paid in recent studies of anaphora to the structural conditions governing anaphora, the basic conception of which problems should to begin with be captured by sentence-level anaphora conditions has, essentially, remained the same since the earliest studies of anaphora in generative linguistics. Anaphora studies still concentrate primarily on the conditions for definite NP anaphora and avoid, in most cases, the problem of the interpretati'on of pronouns (e.g. whether they are interpreted as bound variables or as referential expressions) by postulating notions like coreference, or non-coreference, governed by a coindexing mechanism. However, the coindexing mechanism is a syntactic device which needs to be interpreted semantically, and, similarly, notions like 'coreference' are useful only if there is a clear semantic analysis for them.
