Natural Language Processing Scientific Literature DEMONSTRATIVE ANAPHORA: FORMS AND FUNCTIONS IN FULL-TEXT SCIENTIFIC ARTICLES

This study examines the functions and characteristics of demonstrative anaphora (this, these, that, those) in a collection of full-text scientific documents, confirming that they play an important role in maintaining discourse focus and binding together cohesive sections of text. Unlike corpora in other subject domains, the Cystic Fibrosis database contains more demonstrative expressions than any other class of anaphora. As participants in intersentential reference, demonstratives often refer to complex propositions rather than simple noun phrases. While this tendency complicates automated resolution, our results yield some suggestions toward a resolution algorithm. Primarily, we argue for the incorporation of demonstrative form since different types of demonstratives show different patterns regarding antecedent length and composition. Although further analysis is necessary, our findings provide a groundwork for future exploration.