Determining Subjectivity in Text: The Case of Backward Causal Connectives in Dutch

Abstract

This article addresses the question of how to systematically determine the degree of subjectivity expressed in a text, more specifically, the degree of subjectivity expressed in causal coherence relations. The main hypothesis is that the distribution of Dutch backward causal connectives (want, omdat, aangezien, and doordat) can be explained by the degree of subjectivity of the participant primarily involved in the causal relation (the causally primary participant or CP). The notion of subjectivity is defined on the basis of 2 different conceptions of subjectivity: speaker subjectivity and perspective. On the basis of this notion, a set of 8 text features is established, namely the CP's role in the causality, his or her nature, the mode in which his or her words or thoughts are represented, referential choice (presence and type of referential expression), and predicate features (tense, voice, polarity). This set of features is subsequently applied to an analysis of backward causal connectives in Dutch. The results of a series of corpus analyses are reported. The results show that the distribution of these connectives can successfully be explained in terms of the concept of subjectivity as developed in this article, and they underline the importance of investigating the expression of the self in text.