The Hierarchical Organization of Text as Conceptualized by Rhetorical Structure Theory: A Systemic Functional Perspective

Abstract

Rhetorical Structure Theory (RST) is an approach to the study of text organization which conceptualizes in relational terms a domain within the semantic stratum. In this paper, within an RST conceptualization of text organization, I propose that there are rhetorical subgroupings (or bands) between the local level of text organization, which is grammaticalized by clause complexing, and the upper most level of text organization, which potentially coincides with generic structure. Based on a quantitative analysis, I present a way of viewing text organization in terms of these rhetorical subgroupings within a text that is determined by clause complex boundary and text span size. The corpus used in this study comprises six undergraduate essays that were analysed using RST. In my corpus, other than rhetorical relations that occur within clause complexes and those that occur at the macro level of organization, there are three levels of rhetorical subgroupings of text spans, with identifiable subsets of rhetorical relations. These subgroupings provide a way of conceptualizing mid-level organization within a text, that is the rhetorical ‘chunking’ of different text span sizes within a large text.

Notes

¹The six undergraduate essays were written for a First Year Psychology unit at Macquarie University in 1995 and all had been awarded an A grade. In 1996 Ruth Busbridge and I analysed these essays (as well as six other essays that had been awarded a C grade) using Rhetorical Structure Theory principles. These analyses appear in a report that I wrote, entitled Constructing an Argument in Psychology: Rhetonical Structure Theory and the Analysis of Student Writing, which was part of the ARC funded research project Framing Student Literacy: cross-cultural aspects of English communication skills in Australian university settings (1998). This research was undertaken by the Department of Linguistics and the National Centre for English Language Teaching and Research (NCELTR) at Macquarie University, Sydney, in collaboration with the University of Western Australia, Curtin University and Edith Cowan University.

²In 1991, Mann and Matthiessen had made some preliminary comparisons between the two frameworks in terms of how they both represent ‘the same identification and grouping of functions of language in text’ (p. 244).

³For further discussion of RST, including definitions of the relations, see Mann and Thompson (1987a) and Mann et al. (1992).

⁴The fund-raising letter was used in an edited book as a text that was analysed using different linguistic analyses in each chapter. Consequently the segments it was divided into were broad. The segments were then further subdivided into suitable segments for an RST analysis by Mann et al.

⁵In other published presentations of RST, there is no arc represented for a Joint.

⁶A Joint in the original conceptualization of RST (Mann & Thompson 1986, 1987a, 1987b, 1988 ; Mann et al. 1992) was a type of multi-nuclear configuration that did not involve a relation. So a Joint was a multi-nuclear schema (without a relation), since what in SFL terms is referred to as taxis was referred to as ‘schema’ in the original terminology. Joint has been included in Matthiessen's typology as an extending experiential (additive) relation. See Stuart-Smith (2001) for corpus-based evidence supporting the fact that there is an extending relation between the text spans in a Joint .

⁷The 543 occurrences of relations were in the combined data of the six essays (Essays 1–6). The sample included one complete query (that is, no relation was able to be assigned to the relationship between the text spans) and 16 queries associated with relations (that is, the two analysts considered a particular relation was possible, but had doubts).

⁸The two alternative clusters in the second last subgrouping could, on first glance, make one think that they correspond with a shorter or longer length of essay. However, one of the essays involving text span unit sizes in the low to mid 30s was the shortest essay in the corpus with 77 units and another one was the longest essay in the corpus with 123 units. One could speculate that this difference in clusters indicated a difference in style in text organization or that it could be purely content related. The fact that a variation occurred in the second highest level of text organization is interesting in that it suggests that, after the initial macro structure is established, there follows some possible rhetorical variation in establishing the next level of text organization in this type of text, but that, once that has been established, more uniformity follows.

⁹Queries about relations occurred in all bands except Band 5.

¹⁰The macro-organizational level in an essay includes the top level relation, such as an Elaboration , that links the central nucleus to the body of the essay.

¹¹Other related work (Stuart-Smith 2001) makes a small contribution to the exploration of Band 2.

¹²Field, tenor and mode are the contextual variables involved in the immediate environment in which a particular instance of language is actually occurring. Field relates to what is happening, tenor relates to who is taking part, and mode relates to the role that language plays.

¹³If the mode is ancillary, certain aspects of structure are realized by action and other semiotic systems.

¹⁴However in 2004, Matthiessen proposed another possibility that he suggests needs consideration, namely, whether ‘rhetorical–relational organization may be better reinterpreted as “contextual” rather than semantic’ (2004: 9).