Identity Chains and Noun Phrase Selection: A Case Study of Italian

This study has one main objective, namely the description of the factors in Italian that affect selection of noun phrases or, as we shall call them more specifically, nominal entity referring expressions. To examine such a basic point, we shall have to look carefully at many concepts such as reference and anaphora which, though central to language, have still not been fully described. In particular, we shall have to devise terminology suited to our specific needs even though this may replace more conventional, even traditional, terms. The study is divided into six chapters. The first is a general introduction and deals with cohesion (i.e. the various relations between elements within texts) and, in particular, with coreference. The second looks at reference, that is the relationship between linguistic elements and what they represent. The third chapter presents the solution that we offer, for the purposes of this study, to the problem of defining referring expressions, and also includes a general classification of the various types. The fourth consists of a detailed study of the types of nominal entity referring expression in Italian. In parts, we confirm conventional descriptions of noun phrases and in others we depart from them. Careful examination of types of nominal entity referring expression is, however, necessary because one must look at the similarities and differences between alternatives closely in order to understand the factors affecting the choice of one type rather than another. In the fifth chapter, we discuss the four factors which we believe motivate choice between the various types of nominal entity referring expression. Chapter 6 presents two sets of analysis designed to test the applicability to texts of the factors identified in the previous chapters. The first analysis is statistical and consists of the examination of figures arrived at from a small corpus of randomly selected texts. The second consists of the analysis in context of identity chains in three more randomly selected texts. In our conclusion (6.5), we briefly summarise our findings and argue that the text-oriented approach to noun phrase selection is of value and is an approach that merits further exploration, both in Italian and other languages. We also make some comments on the general significance to text analysis in general of various observations we have made in the course of this study. Finally, we identify related areas in which this study may be of interest, and we highlight areas for further research.