This paper describes corpus-based research on anaphoric relations in spoken Portuguese, relying on data collected in dialogues recorded in real-life situations. The essential analitycal tool is a corpus annotation which classifies each case of anaphora according to four attributes described in the paper. The research project as a whole is concerned with possible applications in natural language processing, particularly regarding natural language interfaces to databases.
Anaphora; Corpus annotation; Corpus linguistics; Natural language processing