From Sentence Structure to Discourse Connectedness

The Praguian notion of the information structure of the sentence (functional sentence perspective, topic-focus articulation, henceforth TFA) as analyzed in the writings of V. Mathesius, J. Firbas, F. Danes, and, within an explicit theoretical framework, in Sgall et al. (1986) has found its counterparts in most different present-day theoretical frameworks and is viewed as an important aspect of the analysis of discourse, contributing to discourse cohesion and to the study of the use of anaphoric pronouns. The dichotomy of topic and focus may be understood as the linguistic patterning basically corresponding to the cognitive opposition between 'given' ('known') and 'new' information, with the caveat that a 'given' item can be referred to as not 'known', esp. if put into contrast. A weak anaphoric pronoun can only be used as contextually bound, i.e. in the topic (unless embedded within the focus); in the prototypical case, it is coreferential with an item occurring in one of the closely preceding utterances, and it always refers to an entity that in the given time-point is above a certain threshold of salience in the stock of shared knowledge (SSK) assumed by the speaker to belong to the hearer's short-term memory; the original model we refer to in our paper was presented first by Hajicová and Vrbová in 1981, and is further developed in Hajicová 1987, Hajicová 1993, and Hajicová 1997. SSK has a dynamic character: the repertoire of elements included in it as well as their degrees of activation (salience) change as the discourse progresses. The difference in the degrees of activation reflects the contention that some elements are relatively easier to access then some others. The decisions on a possibility/impossibility of pronominal reference on one side, and the resolution of anaphora on the other side thus can be based (among other issues, of course) on the difference of the degrees of activation. An element is considered to be activated to a higher or lower degree depending on various factors. In our paper, based on an analysis of discourses of a monologue character, we will discuss some heuristics for the determination of this degree based basically on the topic-focus articulation of sentences; the degrees of activation impose a partial ordering on elements in the stock of shared knowledge, and it will be illustrated that an analysis in this way can also throw an interesting light on the segmentation of the text and on the notion of the 'topic' of the text.