Mann and Thompson (1986, 1988): Rhetorical structure theory Mann and Thompson’s (1986, 1988) and Kintsch and van Dijk’s (1978)

ap-proaches differ considerably from the type of studies discussed so far: instead of investigating particular linguistic elements to explain textual coherence, they use the texts as merely starting points to their analyses and work with the propositions present in them. Mann and Thompson characterize and explore the relations between particular propositions, whereas Kintsch and van Dijk identify the propositions within texts and investigate their semantic structure to describe the system of mental operations that underlie the processes of text comprehension and production. Hence these analyses are not language-specific; they can be applied to any natural language.

According to Mann and Thompson’s (1986) definition, a text is coherent if its parts “go together” (p.58). Its quality of “going together”, however, is not the result of “some” process (e.g., alphabetical sorting of sentences), but it

90 Krisztina Károly

is the consequence of the language user’s ability to impose connectivity on disconnected parts of a visual image. Consider for example the following two sentence pairs:

(1) I love to collect classic automobiles. My favourite car is my 1899 Duryea.

(2) I love to collect classic automobiles. My favourite car is my 1977 Toyota.

Mann and Thompson point out that the second example is incoherent, because while in (1) the implicit proposition is that the instance of the generalization expressed in the first part is represented in the second part, the Duryea is a

“classic automobile”, the Toyota in (2) does not qualify for the same function.

It is not regarded as a typical classic automobile. In (1), the relation between the first and the second sentence is that of elaboration, because the text elabo-rates, or further specifies the concept (“classic automobile”) conveyed by the first part.

In sum, Mann and Thompson claim that relational propositions (RP) arise from two parts of a text, but are not independently derived from either of these parts: they are combinational phenomena, defined on two portions of a text. Thus people’s perception of coherence in a given texts results from the way they can perceive these relations. Mann and Thompson come up with a comprehensive, but not an exhaustive list of RPs. Table 3 provides a summary of these, together with a characteristic example of their occurrence:

Relational

proposition Examples

Solutionhood I’m hungry. Let’s go to the Fuji Gardens.

Evidence They’re having a party again next door. I couldn’t find a parking space.

Justification I’m Officer Krupke. You are under arrest.

Motivation Go jogging with me this afternoon. You’ll be full of energy.

Reason I’m not going to start learning Dutch. You can’t teach an old dog new tricks.

Sequence The huge rod wad released at an altitude of 6 miles. It struck with such force that it buried itself into the ground.

91 Discourse analysis and language teaching

Relational

proposition Examples

Enablement Could you open the door? Here’s the key.

Elaboration I love to collect classic automobiles. My favourite car is my 1899 Duryea

Restatement He sure beat me up. I really took a thrashing from him.

Condition Slowly stir the powder into the fluid. The mixture will be very thick.

Circumstance I went hitchhiking in Norway. Nobody would pick me up.

Cause There were landslides in Malibu last week. Four neighborhoods lost their electricity.

Concession I know you are great credentials. You don’t fit the job description because this job requires someone with extensive experience.

Background Hayes just resigned. He’s our chancellor.

Thesis-Antithesis This book claims to be a guide to all the trees of Indiana. It’s so incomplete that it doesn’t even have oak trees in it.

Table 3: Summary of relational propositions

Describing RPs, Mann and Thompson identify five main characteristics which generally apply to them. First, they claim that RPs are “basic.” This means that while other inferences depend on RPs, they do not depend on any other inferences, in other words, RPs lead inferentially to other propositions.

Second, RPs arise independently of any specific signals of their existence:

in some cases they might be indicated by explicit linguistic material (e.g., conjunctions, subordinators), but they do not need to be signaled in order to exist. What is important is that they are implicit relations, therefore they may be conveyed even without explicit markers. Third, RPs are involved in communicative acts, since they are communicated with the illocutionary force of assertions, questions, or commands. Fourth, RPs are not limited to organizational aspects of texts. They are inherently combinatorial, and thus do more than simply relate parts of text: they are involved deeply in relat-ing subject-matter-specific conceptions to each other, and, consequently, convey essential subject matter. The final characteristic of RPs is that they

92 Krisztina Károly

are essential to the effective functioning of a text, because without them the text lacks coherence.

Based on these principles, Mann and Thompson (1988) created a functional theory of text organization, referred to as Rhetorical Structure Theory (RST).

RST is used to identify the hierarchic structure of texts through the investiga-tion of RPs (i.e., unstated but inferred proposiinvestiga-tions that arise from the text structure in the process of interpreting texts). It describes the relations between text parts (clauses) in functional terms, identifying the transition point of a relation and the extent of the items related.

The relations (Table 3) are defined to hold between two non-overlapping text spans, here called the nucleus and the satellite, and they produce pat-terns which are called schemas. Schemas define the structural constituency arrangements of text: “they are abstract patterns consisting of a small number of constituent text spans, a specification of how certain spans (nuclei) are related to the whole collection” (p.247). In other words, they determine the possible RST text structures.

Based on Mann and Thompson’s analyses, it may be assumed that RPs are essential to the coherence of their text. Recognizing these relations (i.e., con-structing their RST structure) leads to the actual understanding of the text.

It is typical, but not universal for texts to be hierarchically structured and functionally organized. Thus not all texts have RST structure, but potentially every text can have one. In the Anglo-Saxon culture the predominant mode for texts is to have a RST analysis. Examples of these are, for instance, maga-zine articles, letters, essays, or scientific abstracts. Therefore RST provides a substantial basis for the study of coherence in numerous genres.