Definition:
Rhetorical structure refers to the functional and hierarchical organization of discourse — how segments of a text (clauses, sentences, paragraphs) are connected through non-arbitrary relationships that together realize the communicative purpose of the whole. The concept is most precisely formalized in Rhetorical Structure Theory (RST), developed by Mann and Thompson (1988), which provides a systematic taxonomy of relations that can hold between text spans — including elaboration, contrast, cause, evidence, concession, background, and purpose — and proposes that well-formed texts have a hierarchical structure in which every span participates in a rhetorical relation to adjacent and higher-level spans.
Core Concepts of RST
Nucleus and Satellite:
Most RST relations are asymmetric:
- Nucleus: the more essential span — the core information the writer intends to convey
- Satellite: a supporting span that elaborates, motivates, or conditions the nucleus
Example (Evidence relation):
> [The hypothesis is well supported.](Nucleus) [Three independent experiments found the same result.](Satellite)
The satellite provides evidence for the nucleus; if the satellite is removed, the core communicative intent survives (impoverished but intact).
Common RST Relations
| Relation | Description |
|---|---|
| Elaboration | Satellite provides more detail about nucleus |
| Background | Satellite provides context for understanding nucleus |
| Motivation | Satellite increases the reader’s desire to do the nucleus action |
| Evidence | Satellite supports belief in nucleus |
| Concession | Writer acknowledges something contrary to nucleus but maintains nucleus |
| Contrast | Nucleus and satellite are compared as alternatives |
| Cause | Satellite is the cause of nucleus situation |
RST Trees
RST analyses produce tree structures:
- Terminal nodes are text spans (typically clauses or sentences)
- Internal nodes represent rhetorical relations
- The top of the tree is the overall communicative purpose of the text
A coherent text has a single tree; a text lacking overall purposive unity may fail to produce a coherent RST analysis.
RST and Text Coherence
RST provides a formal account of textual coherence — how a collection of sentences becomes a unified text rather than random sentences. Incoherent texts fail to establish RST relations between their parts.
RST in SLA Writing Research
RST has been applied to analyze:
- Cross-cultural differences in argumentative organization (Arabic vs. English, Chinese vs. English)
- L2 learner texts showing less hierarchically integrated rhetorical structure than L1 academic texts
- The development of argumentative organization in L2 academic writing across proficiency levels
History
William Mann and Sandra Thompson developed RST at USC Information Sciences Institute in the 1980s, publishing the foundational paper in 1988. RST was originally developed for computational linguistics (natural language generation) but has been extensively adopted in discourse linguistics, writing research, and NLP. RST annotation schemes are used in computational discourse parsing.
Common Misconceptions
- “Rhetorical structure = paragraph organization (intro-body-conclusion).” Rhetorical structure operates at all levels simultaneously — from clause combinations to whole-text organization — not just at the paragraph level.
- “All texts have the same rhetorical structure.” Different text types and genres have different characteristic rhetorical structure patterns; RST describes what is there, not what should be.
Criticisms
RST has been criticized for: (1) being difficult to apply consistently — annotators often disagree on relation labels; (2) the nucleus/satellite asymmetry not holding for all relations (some are multi-nuclear); (3) imposing a tree structure that may not reflect real discourse organization, which may be more network-like. Despite these criticisms, RST remains the most widely used formal discourse structure framework.
Social Media Sentiment
Rhetorical structure theory is primarily academic but appears in NLP and computational linguistics communities where discourse parsing is an active research area. RST parsing is used in summarization, information retrieval, and essay scoring systems — connections that give it practical computational relevance beyond theoretical discourse linguistics.
Last updated: 2025-07
Practical Application
For L2 writing teachers, the RST framework provides a principled tool for analyzing student argumentation: identifying where relations are unclear, implicit, or logically unconnected helps target feedback beyond surface-level style. Students can be taught to make rhetorical relations explicit through connectives and to check whether their text segments logically connect.
Related Terms
See Also
Research
Mann, W. C., & Thompson, S. A. (1988). Rhetorical Structure Theory: Toward a functional theory of text organization. Text, 8(3), 243–281.
The foundational RST paper, introducing the full taxonomy of relations, the nucleus-satellite distinction, and the tree representation of discourse structure — essential reading for anyone working with formal discourse structure analysis.
Taboada, M., & Mann, W. C. (2006). Rhetorical Structure Theory: Looking back and moving ahead. Discourse & Society, 17(3), 423–452.
A comprehensive retrospective and forward-looking assessment of RST, covering the theory’s development, applications in NLP and linguistics, and the outstanding challenges — the best overview of the field.
Connor, U. (1996). Contrastive Rhetoric: Cross-Cultural Aspects of Second Language Writing. Cambridge University Press.
An application of discourse structure analysis (including RST-informed approaches) to contrastive rhetoric — examining how cultural and linguistic backgrounds shape L2 writers’ argumentative organization across languages and cultures.