Control Issues in Anaphor Resolution

Much research in computational linguistics has concentrated on treating the individual phenomena of natural language rather than on how these treatments can be made to work together. This paper discusses four respects in which processes embodying individual treatments must be made to cooperate with one another if anaphoric expressions are to be interpreted correctly. These are the coordination of (1) treatments of anaphoric and non-anaphoric ambiguity in a sentence; (2) treatments of diierent kinds of knowledge relevant to interpretation; (3) the resolution of several anaphors in one sentence; and (4) candidate referents arising from several sources. It is argued that a exible control structure based on numerical scoring allows the required cooperation to take place, whereas a more limited depth-rst architecture seems not to allow this. The discussion is grounded in comparisons between two very diierent implemented systems that resolve anaphors. The rst, SPAR, is an experimental system embodying the control structure argued for. The second, the SRI Core Language Engine, is a wide-coverage system intended for practical applications. Its control structure is, at the time of writing, depth-rst, but planned enhancements will allow the problems discussed in the paper to be tackled, resulting in the development of a powerful and widely-applicable natural language processor.