Argumentative text as rhetorical structure An application of rhetorical structure theory.pdf

Argumentative Text as Rhetorical Structure:
An Application of Rhetorical Structure Theory
M. AZAR
University of Haifa
Hebrew Language Department
31905 Haifa, Israel
ABSRACT: Rhetorical Structure Theory (RST), as a tool for analyzing written texts, is par-
ticularly appropriate for analyzing argumentative texts. The distinction that RST makes
between the part of a text that realizes the primary goal of the writer, termed nucleus, and
the part that provides supplementary material, termed satellite, is crucial for the analysis of
argumentative texts.
The paper commences by determining the concept of argument relation (argument + con-
clusion) and by briefly presenting RST. It continues by identifying five of RST’s rhetorical
relations of the satellite/nucleus schema (Evidence, Motivation, Justify, Antithesis,
Concession) as five argument relations, each being, logically or pragmatically, a special kind
of argument: Evidence being a supportive argument, Motivation an incentive argument,
Justify a justifier argument, and Antithesis and Concession persuader arguments. To illus-
trate, an analysis of three short texts concludes the paper.
KEY WORDS: Argument relation, argumentative text, persuasion, rhetorical relation, rhetor-
ical structure, text analysis, writer’s intention
1. INTRODUCTION
Studies on argumentative text structure refer predominantly, directly or
indirectly, to Toulmin’s work (Toulmin, 1958). Toulmin’s model in its
reduced relationship form of Datum or Argument → Conclusion authorized
by a Warrant or Topos was considered the ‘prototypical argumentative
schema’ and used to define a minimal formalization of argumentative text.
Doubts, however, have recently been raised regarding the relevancy of any
prototypical argumentative text schema and especially Toulmin’s schema
(Brassart, 1996a; 1996b). I share these doubts, and think that argumenta-
tive text should be analyzed first of all according to a general theory of
text analysis, and also be recognized as argumentative by criteria pertaining
to that theory. All other examinations concerning the specifics of the
argumentation found in the text are theory-dependent, and may find their
place in addition to the text analysis, which is (or should be) independent
of any specific argumentation theory. The text theory I recommend for a
preliminary analysis, which can be followed by a more specific argumen-
tation analysis and theory dependent, is Rhetorical Structure Theory (RST).
RST was conceived originally for a text generation program, and was
ARGU ART. NO. 117 PIPS. NO. 182288
DISK, CP
Argumentation 13: 97–114, 1999.
 1999 Kluwer Academic Publishers. Printed in the Netherlands.

used in a large number of natural language generation systems (for example,
Hovy, 1991; Knott, 1991; Moore and Paris, 1989; Rosner and Stede, 1992).
It was also used in many descriptive studies of discourse (Fox, 1987; Linden
et al., 1992; Matthiessen and Thompson, 1988; Abelen, Redeker and
Thompson, 1993), as well as in a case-study to facilitate textlinguistic com-
parison of two texts and accurate diagnosis of the sources of the incoher-
ence of one of them (O’Brient, 1995).
Mann and Matthiessen (1991, p. 235) state that distinguishing between
the nucleus part of a statement, i.e., the part that relates to the primary
goal of the writer, and the satellite part, which realizes the writer’s goal
and provides supplementary material, is ‘crucial to a study of persuasion
texts’, and therefore, RST is a useful tool for the analysis of persuasive
texts. Abelen, Redeker and Thompson (1993, p. 333) also see RST as
‘especially useful for the analysis of persuasive language, because it allows,
and even forces, the analyst to consider the intended communicative effects
expressed in, or plausibly inferable, from the text.’
It is true that certain difficulties are encountered at times in making a
distinction between a nucleus and a satellite (see Martin, 1992, pp.
253–265), and no fixed, one-to-one mapping between intentional and
informational levels exists (see Moore and Pollack, 1992), but these critics
are less relevant for the analysis of texts which intend to persuade by using
the form of argument relations, (i.e., a reason or some other kind of
argument and a conclusion), because these argument relations, which are,
as we will see, nothing else than certain types of satellite/nucleus relations,
are recognized as such independently of the difficulties that may exist
in identifying other satellite/nucleus relations that may take place in the
text. The first kind of satellite/nucleus relations, the one which is of our
interest here, is by definition intentional (or presentational in RST’s
terminology), and as such, the distinction between nucleus and satellite
is indispensable, and no difficulties are encountered in making the dis-
tinction. The mapping between intentional and informational (subject matter
in RST’s terminology) should not concern us, because all the intentional
relations are, by definition, independent of any specific type of informa-
tional relation.1
In a previous paper (Azar, 1995) I contended that five of RST’s rhetor-
ical relations – Evidence, Justify, Motivation, Antithesis and Concession
– are congruent with the concept of argument relation, and revealed that
Evidence, Antithesis and Concession are often found in newspapers in the
form of texts (or parts of texts) built in two hierarchical levels, each con-
taining two units connected by one of the three relations, all the six theo-
retically possible combinations have been found and exposed. My aim in
the present paper is twofold. Firstly, there is a need to clarify what is exactly
meant by argument relation, and to justify why the five RST relations men-
tioned above should be considered as argument relations. Secondly, and
this is my principal task, to demonstrate how RST can be useful in iden-
98 M. AZAR

tifying and analyzing argumentative texts and differentiating them from
explanative texts.
Firstly, I will explain my conception of argument relation, and will then
briefly introduce RST, quoting the five RST definitions of the five rela-
tions that I consider as argument relations. Secondly, I shall discuss the jus-
tification of seeing them congruent with the notion of argument relation,
and give some examples of analysis.
2. WHAT IS AN ARGUMENT RELATION?
I entirely agree with Connor and Lauer (1985), who are in sharp contra-
diction to Ducrot (1992), in maintaining that argumentation is a type of
persuasive activity, which, in accordance with long rhetoric tradition, inte-
grates the rational, credibility, and affective appeals corresponding to logos,
ethos, and pathos. Understanding argumentation as a persuasive activity is
congruent with the view of Chittleborough and Newman (1993, p. 202)
who contend that
An argument has been put forward where there has been an intention to either establish
a proposition, or persuade one or more people to accept a proposition (where such an
acceptance would involve a change in belief, strength of belief, or a change in behav-
iour).
Chittleborough and Newman add that the object-matter of an argument’s
intention (to establish or persuade) is its conclusion, which can be a pre-
scribed action or an assertion, and this attempt to achieve an argument’s
intention will involve the use of at least one supportive and/or at least one
persuader. A supportive is defined (p. 196) as being a reason or item of
information presented in an argument which is intended to provide ‘support
for a conclusion,’ and a persuader as ‘a psychologically manipulative tech-
nique used by an arguer with the intention or hope of increasing the chances
of the conclusion being accepted by a recipient.’ An ‘item of information’
may include visual images such as charts, diagrams, graphs, symbols,
pictures, photos, etc.
Knowing now what is meant by argument, supportive, persuader, and
conclusion, we may formulate a (first) definition of the term argument
relation:
An argument relation is the relationship that exists between two parts of a monologue,
one being an argument (i.e., a supportive or a persuader), the other a conclusion (i.e., a
prescribed action or an assertion).
This definition of argument relation is compatible with most contempo-
rary argumentation scholars, for example, Van Eemeren et al. (1987),
Brockriede (1990), Crosswhite (1996) – to mention just a few. Perelman
(1982, p. 47), in particular, makes a very clear distinction between ‘argu-
mentation’ (i.e., our argument relation) and ‘demonstration’:
ARGUMENTATIVE TEXT AS RHETORICAL STRUCTURE 99

Argumentation differs from formal demonstration in three important respects. First
demonstration is possible only within a formal system of axioms and rules of inference.
Argumentation does not start from axioms but from opinion, values, or contestable
viewpoints; it makes use of logical inferences but is not exhausted in deductive systems
of formal statements. Second, a demonstration is designed to convince anybody who has
the requisite technical knowledge, while argumentation is always directed to a particular
audience and attempts to elicit or increase the adherence of the members of the audience
to the theses that are presented for their consent. Finally, argumentation does not aim
at gaining purely intellectual agreement but at inciting action, or at least at creating a
disposition to act at the appropriate moment.
Perelman’s ‘demonstration’ does not correspond to Chittleborough and
Newman’s ‘establishing a proposition’, because the latter does not need ‘a
formal system of axioms and rules of inference’. It needs only a supportive,
which is a rational persuasion argument as defined by Blair (1992, p. 258):
In general terms, argumentative discourse is considered rational persuasion if it consists
of a (set of) proposition(s), advanced as a reason for accepting another proposition
or for performing an action which are intended to be so related to it that it would be
inconsistent (in some sense) to accept the (set of) reason(s) but not accept the proposi-
tion or endorse the performance of the action in question.
A persuader, on the other hand, is the kind of argument that can be accepted
without being inconsistent if the conclusion is not acceptable. Indeed, in
real discourse, people often use non-reason persuader arguments, which are
not intended to be, nor intended to appear as, reasons or a rational argu-
ments, but are nevertheless put forward with the intention or hope of
increasing the chances of the conclusion being accepted by a recipient. In
Azar (1995) and in the present paper, it is shown that antithesis and
concession expressions can be used as argument relations of the persuader
type. We will also see that taking RST’s definitions in consideration, will
force us to enlarge the notions of argument (or persuader).
3. RHETORICAL STRUCTURE THEORY (RST) AND ARGUMENT RELATIONS
Rhetorical Structure Theory (RST) is a theory of text organization. It
describes the kinds of parts a text can have, how they can be arranged and
connected to form a whole text. The authors of RST (Mann and Thompson,
1986; Mann and Thompson, 1987; Thompson and Mann, 1987; Mann and
Thompson, 1988; Mann et al., 1992), assert that except for certain types
of text, such as laws, contracts, reports and various kinds of language-as-
art (poetry), virtually every coherent text has an RST analysis. Relationships
between parts of a text, they maintain, are crucial to make the text function
as a single unit. The theory maintains that consecutive discourse elements,
termed text spans, which can be in the form of clauses, sentences, or
units larger than sentences, are related by a relatively small set (20–25) of
rhetorical relations. Mann and Thompson (1988) present a list (not exhaus-
tive) of these relations with their definitions. According to RST, a text can
100 M. AZAR

generally be described by breaking it down into two spans, and the rela-
tionship, termed schema, between them is mostly asymmetric: the one,
termed nucleus, is usually more essential to the text than the other, termed
satellite. Mann and Thompson (1988) maintain that the removal of text
spans which function only as satellites, and never as nuclei, will leave a
coherent text with a message resembling that of the original, whereas the
same operation exercised on the nuclei will result in an incoherent text
whose central message is difficult or impossible to comprehend.
The definitions of the relations is made with reference to two fields,
specifying a particular judgment the text analyst must make in building
the RST structure. The first field includes constraints on each one of the
two text spans and on their combination, the second specifies the plau-
sible effect that the writer attempts to achieve on the reader’s belief,
inclination, or understanding. The locus of the effect, derived from the
effect field, is identified as either the nucleus alone or the nucleus-satel-
lite combination. Thus RST provides a set of the writer’s intentions and
the conditions which enable the reader or analyst to identify those inten-
tions. The task of the analyst is to break the text down into text spans and
to find a RST relation that connects each pair of spans until all pairs are
accounted for. To determine whether or not a relation holds between two
spans of text, the analyst examines whether the constraints on the nucleus
and satellite hold and if it is plausible that the writer’s point has the desired
effect on the reader.
Mann and Thompson (1988, p. 257) divide RST relations into two
classes: subject matter and presentational. Subject matter relations are those
whose intended effect is that the reader recognizes the relation in question:
presentational relations are those whose intended effect is to increase some
inclination in the reader, and the locus of the effect is the nucleus alone.
Mann and Thompson (1988) list sixteen subject matter relations and seven
presentational relations as shown below:
Subject Matter: Elaboration, Circumstance, Solutionhood, Volitional Cause, Volitional
Result, Non-Volitional Cause, Non-Volitional Result, Purpose, Condition, Otherwise,
Interpretation, Evaluation, Restatement, Summary, Sequence, Contrast.
Presentational: Motivation (increases desire), Antithesis (increases positive regard),
Background (increases ability), Enablement (increases ability), Evidence (increases
belief), Justify (increases acceptance), Concession2
(increases positive regard).3
The most important difference between the two groups of relations is
that in presentational relations, ‘the mapping between intention and rhetor-
ical relations is a one-to-one mapping’ (Moore and Paris, 1993, p. 667).
For example, if the writer’s goal is to increase the reader’s desire to perform
the action presented in the nucleus, then whatever text is used to achieve
this goal, it stands in a Motivation relation to the nucleus (see below (b)
the definition of Motivation). On the other hand, if the writer’s goal is only
to inform the reader of what has happened, without intending to affect his

belief or acts, then the relation must be chosen according to the informa-
tional link that holds between the text spans. For example, if the writer’s
goal is to inform the reader that a scheduled soccer game will not take place
if it rains, this information will be presented by using the relation called
Condition, and not by any other relation. However, not every conditional
sentence, in the syntactical sense of the term, is a Condition relation in RST
sense; it can also be, for example, a Motivation relation. Thus, the sentence
If you decide to buy it now, you’ll get a reduction of 25 percent
(which can be found in advertising) may well be analyzed according to
RST as a text composed of two minimal text spans associated by the
relation called Motivation.; the subordinate if-clause being the nucleus, and
the consequent (main) clause the satellite. Figure 1 below shows this
analysis by using a RST diagram:
Figure 1. RST Diagram of the text (1) If you decide to buy it now, (2) you’ll get a reduc-
tion. (The vertical line indicates that the text span below is the nucleus; the curved line
indicates the existence of a nucleus/satellite relation between the two text spans.)
This analysis accounts not for the information (that is, RST’s subject
matter) conveyed by the clausal components of the discourse, but for what
is inferred, namely, the writer’s intention to increase (to ‘motivate’) the
reader’s desire to buy the merchandise by offering him or her a price reduc-
tion. The analyst, it must be emphasized, cannot identify the rhetorical rela-
tions in a text by relying on forms, connectives or discourse markers; he/she
must essentially rely on context and situation in order to grasp the writer’s
intention. Unfortunately, RST does not provide a theory or a methodology
to make the ‘correct’ RST analysis, but the authors of RST insist (see
specifically Mann and Thompson, 1986) that as a simple matter of fact,
we do recognize the relation holding between two text segments indepen-
dently of any specific signal of its existence. In their view, with whom I
entirely agree, the linguistic signals do not actually ‘signal’ relations in any
direct way. A more appropriate description of their function would be that
they constrain the interpretation of the relations between text spans in a
rather loosely way, since the relationship between those ‘signals’ and
semantic/rhetorical relations is not one-to-one, but many-to-many. The
102 M. AZAR

authors’ conclusion is (Mann and Thompson, 1986, pp. 71–72): ‘but we
do not believe that there are undiscovered signal forms, and we do not
believe that text form can ever provide a definitive base for describing how
relational propositions can be discerned.’ Only functional, i.e., semantic/
pragmatic characterization, can be definitive, albeit depending on an inter-
pretation which is not necessarily exclusive.
We will now examine which of the seven presentational relations listed
above can be considered argument relations, i.e., the satellite being an
argument, and the nucleus an opinion/conclusion. It is perfectly clear that
since any argument, by definition, is always intended to increase belief or
disposition to act, no subject matter relation can be considered an argument
relation because their intended effect is to recognize, not to increase. Also,
the raison d’être of an argument is to advance something other than itself,
namely, a conclusion, therefore, only those relations whose locus of effect
is the nucleus alone should be considered as possible argument relations,
and all the seven presentational relations have this property.
Of the seven relations that are intended to increase, that is, to increase
(a) desire to act (Motivation), (b) positive regard (Antithesis and
Concession), (c) ability to understand (Background), (d) ability to act
(Enablement), (e) belief (Evidence), and (f) readiness to accept (Justify) –
two have not the kind of effects that are congruent with those expected
from an argument relation: increasing ability to understand (c) and
increasing ability to act (d). Increasing ability to understand can be con-
sidered as an explanation relation, but not as an argument relation, and
increasing ability to act is not the same as increasing the desire to act.
Background and Enablement, then, which increase ability, but not desire,
should be omitted from our list of possible argument relations. We are left
with five RST relations, whose definitions will be examined in order to
decide their status as argument relations. The five definitions below are
taken from Mann and Thompson (1988), but the examples are chiefly mine
(N = Nuclear; S = Satellite; R = Reader; W = Writer).4
a) Evidence:
Constraints on N: R might not believe N to a degree satisfactory to W.
Constraints on S: The reader believes S or will find it credible.
Constraints on the N + S combination: R’s comprehending S increases R’s belief of N.
The effect: R’s belief of N is increased.
Locus of the effect: N.
Example:
1. The program as published for the calendar year 1980 really does work.
2. In only a few minutes, I entered all the figures from my1980 tax return and got a result
which agreed with my manual calculation to the penny.
Evidence is surely an argument relation, since the intended effect is to
increase belief in what is said in N (the conclusion), and the way to do it
is to put forward an S (an argument) acceptable by the reader (constraints

on S). This is exactly what argumentation is all about: to present an
argument (unit 2 in our example) in order to evidence the truthfulness of
the opinion/conclusion (unit 1) so that the addressee’s (R’s) belief of it
increases. The S of Evidence must be a supportive (not a persuader), since
the constraints on S + N say that ‘R’s comprehending S increases R’s belief
of N’, meaning that if R is considered as a rational person, it is reasonable
for W to believe that if S is presented before R and R understands its
content, he/she will be inclined to accept the truthfulness of N.
b) Motivation:
Constraints on N: Presents an action in which R is the actor (including accepting an offer),
unrealized with respect to the context of N.
Constraints on the N + S combination: Comprehending S increases R’s desire to perform
an action presented in N.
The effect: R’s desire to perform the action presented in N is increased.
Example:
1. Invest $100,000 U.S.
2. and get the Green Card Guaranteed.
The difference between Evidence and Motivation is clear: The authors
of RST think that the goal of increasing R’s desire to perform a certain
action is different from the goal of increasing belief and decided to see
them as two different rhetorical relations deserved to have two different
definitions. The difference does not concern only the goal (N, the conclu-
sion) but also the means to achieve it (S). The absence of constraints on S
in the above definition of Motivation indicates that unlike Evidence, every-
thing goes for achieving the goal of increasing R’s desire to perform the
action. In our example, the second clause is the satellite, and it is more of
an incentive than a reason (a supportive) or a psychological manipulation
(a persuader). The term Motivation is appropriate because it excludes
intimidation and threat, which are not intended to increase desire, and there-
fore, should not be considered as satellite of Motivation.
Considered from the practice of argumentation, one can say that through
Motivation relation (found most of time in commercial advertising), a writer
acts on the concrete actions (such as buying something) of an addressee
by means of an incentive. If I am right in considering an incentive as a
special kind of a persuader, we may add a third kind of argument to sup-
portive and persuader: incentive.
c) Justify:
Constraints on N: None.
Constraints on S: None.
Constraints on the N + S combination: R’s comprehending S increases R’s readiness to
accept W’s right to present N.
The effect: R’s readiness to accept W’s right to present N is increased.
104 M. AZAR

Example:
1. I am Officer Morgan.
2. Please show me your ID.
Increasing readiness to accept the right to present N means nothing else
than increasing readiness to accept the right to perform the speech act of
N. Since no constraints are imposed on S or N, the speech act involved in
Justify can be any speech act. In our example, the speech act of the nucleus
2 is a demand or incitement to perform an action, which could have been,
in another context, a nucleus of Motivation, but in our example, sentence
2 is the nucleus of a Justify relation because sentence 1 is understood as
justifying the unusual specific demand. In another context, the same
sentence 1 can be used to justify the presentation of an assertion:
1. I am officer Morgan.
2. I saw this man putting something in his pocket.
The satellite intended to increase readiness to accept the right to perform
the speech act of N must be considered as an argument, since its ultimate
raison d’être is to convince the addressee to perform the requested act, to
pay attention to the assertion, etc. Therefore, it should be considered as a
fourth kind of argument, which may termed justifier.
d) Antithesis:
Constraints on N: W has positive regard for the situation presented in N.
Constraints on S: None.
Constraints on the N + S combination: The situation presented in N and S are contrasting
(i.e., are (a) comprehended as the same in many respects, (b) comprehended as
different in a few respects and (c) compared with respect to one or more of these
differences); because of an incompatibility that arises from the contrast, one cannot
have positive regard for both the situations presented in N and S; comprehending S
and the incompatibility between the situations presented in N and S increases R’s
positive regard for the situation presented in N.
The effect: R’s positive regard for N is increased.
Example:
1. When on the road, don’t be right,
2. be wise.
e) Concession:
Constraints on N: W has positive regard for the situation presented in N.
Constraints on S: W is not claiming that the situation presented in S doesn’t hold.
Constraints on the N + S combination: W acknowledges a potential or apparent
incompatibility between the situations presented in N and S; W regards the situations
presented in N and S as compatible; recognizing that the compatibility between the
situations presented in N and S increases R’s positive regard for the situation pre-
sented in N.
The effect: R’s positive regard for N is increased .
Locus of the effect: N.5

Example:
1. Although it is toxic to certain animals,
2. evidence is lacking that it has any serious long-term effect on human beings.
One common characteristic of the three first RST relations is that each
one has a very specific intended effect (to increase belief, to increase desire
to act, and to increase readiness to accept the writer’s right to present N).
In order for the writer to achieve those effects, he/she must use a specific
kind of argument (satellite): a supportive for Evidence, an incentive for
Motivation, and a justifier for Justify. Antithesis and Concession , by
contrast, have a vague intended effect of increasing positive regard, and
in order for the writer to achieve this effect, no specific kind of argument
with a special content is needed. What is even more unusual about these
two rhetorical relations is that before being recognized as rhetorical (RST)
relations they must be first recognized as semantic (if not syntactic) rela-
tions, and only after accomplishing this task the analyst may go further and
see if all the conditions of the rhetorical relations take place. In the case
of Antithesis, a contrastive relation must hold, and the incompatibility
existing between N (a denied thesis) and S (the asserted anti-thesis) rhetor-
ically functions as a means to increase positive regard toward the nucleus,
provided that R understands that W has positive regard toward N. In the
case of Concession also, R must not only recognize the apparent incom-
patibility existing between N and S, but also the positive regard that W
has toward N.
The satellite parts of Concession and Antithesis are certainly not sup-
portive (let alone incentives and justifiers); they are persuaders. One should
not be surprised to find these two relations serving in argumentative texts.6
According to Robrieux (1993, p. 73) the effectiveness of a concession,
known in classical rhetoric as argumentative (or rhetorical) prolepsis, lies
in the writer’s presentation of a counter-argument: when using a conces-
sion, the writer states in advance what may be seen as an unfavorable
argument for his belief; by such a maneuver he/she eliminates a possible
unfavorable intervention and also reinforces the credibility of the nucleus,
because the readers are led to understand that the writer has already con-
sidered all the possible objections, or at least the important ones, and
rejected them all as valid counter-arguments. The effectiveness of an
antithesis may result from thinking along the same lines: if the writer rejects
a thesis and replaces it by an antithesis, it is reasonable to expect that he/she
has arrived at this point after having examined all possible alternatives
and rejected all of them. These persuasive techniques are certainly psy-
chologically manipulative techniques, i.e., the two relations are argument
relations, and their argument parts are persuaders.7
We may now reformulate our definition for argument relation:
An argument relation is the relationship that exist between two parts of a monologue, one
being an argument (i.e., a supportive, an incentive, a justifier, or a persuader), the other
a conclusion (i.e., a prescribed action or an assertion).
106 M. AZAR

It is perhaps difficult to accept the idea of viewing incentives (Motivation
satellites), justifiers (Justify satellites) and/or persuaders (Antithesis and
Concession satellites) as arguments, but the point is that these satellites
share something very significant with supportive satellites (i.e., ‘real’
arguments): they all intend to create or to increase positive regard (see
Thompson and Mann, 1987, p. 365), in other words, through them, a writer
acts on the opinions, attitudes, or behaviors of an addressee or audience
by lending credibility or acceptability to a conclusion (nucleus). The inten-
tion to increase positive regard is, in my opinion, a necessary and suffi-
cient condition to consider the five rhetorical relations as argument
relations.
We are now in a position to illustrate the usefulness of RST in analyzing
some small texts and decide if they are argumentative texts and, if so,
what kinds of argument relations they are exhibiting.
4. RST ANALYSIS FROM AN ARGUMENTATION PERSPECTIVE
The first two texts are ‘To-the-Editor Letters’, which appeared in
Newsweek, May 20, 1996, p. 16. Both letters dealt with the same subject:
the high cost of college education. I consider one letter to be a non-argu-
mentative text despite its argumentative appearance, whilst the other I
consider to be an argumentative text in spite of its non-argumentative
appearance.
Text 1 (broken into minimal (i.e., clauses), or nearly minimal text spans):8
1. Newsweek’s otherwise fair and balanced story of the cost of college education is
marred by sensationalism of the cover, which proclaims ‘$1,000 a week: The Scary
Cost of College’ (‘Those Scary College Costs’, Society, April 29).
2. Although you point out in your story that the actual cost at the average college is
far lower than $1,000 a week,
3. public-opinion polls show that the Americans think college costs more than it really
does.
4a. They think this because the media,
5. in their quest for the dramatic and startling,
4b. always focus on the most expensive colleges.
6. By doing so, they mislead the public and distort the situation facing the majority of
students and their families.
Terry W. Hartle
Vice President, American Council on Education
Washington, D.C.
Judging by the general tone of the letter which expresses disagreement
with the magazine’s sensationalism on its cover presentation of the subject,
and the use of connectives which may be found in argumentation (although,
because, in their quest, by doing so) one may think that this text is
argumentative, i.e., it contains some argument relations, if not at its top
hierarchical level, at least in a lower level of the structure. However, if we

ask ourselves which text span should be considered as the nucleus of the
text, i.e. the principal message which the remainder of the text depends on
communicatively, our answer should be: text span 1, with the remaining
text spans 2–6 being its satellite. Our second question, as argumentation
analysts, is: is the RST relation holding between the two text spans one of
the five relation that have been recognized (in the previous section) as
argument relations? For obvious reasons, it is not Antithesis, Concession
or Motivation. It is also not Justify, since the writer surely does not think
that he/she must increase the readers’ readiness to accept his right to present
what he/she wants to say: in Western context of writing letters to the Editor,
it is taken for granted that the writer has the right to present what he/she
has to say. Neither is the relation Evidence, because the writer’s purpose
is not to make the magazine’s Editor or the readers believe that the ‘story
of the cost of college education is marred by sensationalism of the cover.’
Saying that something is ‘marred by sensationalism’ is not an opinion/
conclusion, which the writer wants the readers to be convinced of its
truthfulness, but rather a blame, a criticism,9
which needed to be explained
in order to make the readers understand the causes that brought this blame.
In my opinion, it is an Elaboration relation, because the primary goal of
this satellite part of the text is to explain, or to make understand, the blame
expressed in the nucleus by providing more details. According to Wyatt
and Badger (1993), an expositional composition operates mainly through
logical and explanatory devices to provide understanding of its subjects.
Indeed, text span 2–6 aims at explaining in a rational way the writer’s
assertion that the headline ‘$1,000 a week: The Scary Cost of College’ is
sensationalism.
Text span 2–6 can also be broken down into two parts: 2–3 (N) and 4a–6
(S). The relation between 2-3 and 4a-6 is Non-volitional cause,10
which in
our text, serves as an explanation to the fact presented in 3. Within 2–3,
the syntactic and semantic relation between the two units is concession,
but, as has already been said, RST analysis is not equal to syntactical
analysis. The relation is not RST Concession, because the intended effect
of satellite 2 is not to increase the reader’s positive regard toward nucleus
3, since 3 states an incontestable fact, and facts, if accepted as such, need
no support or ‘positive regard’ of any kind, only explanation. If, despite
its formal appearance, the relation is not Concession,11
it can be
Circumstance, which its effect is that ‘R recognizes that the situation pre-
sented in S provides the framework for interpreting N’ (Mann and
Thompson, 1988, p. 272). In our text, 2 provides the framework for inter-
preting N as standing in opposition against something else: public-opinion
concerning college costs against actual college costs, and this is the reason
of using ‘Although’, to point at the incompatibility between what ‘you point
out in your story’ and, in spite of that, what ‘the Americans think.’ By
adding the concessive clause, the writer gains a more relevant, interesting,
and surprising effect of the information presented in N.
108 M. AZAR

Text span 4a–6 can be split into 4a–4b (N) and 6 (S), exhibiting a Non-
volitional result relation.12
Finally, 4a–4b breaks into 4a + 4b (N) and 5
(S), and this is a Volitional cause relation.13
We can now present the
rhetorical structure of text 1 (Figure 2 below) using a RST diagram.
Figure 2. RST diagram of Text 1.
The diagram of Text 1 illustrates, among other things, that the text contains
no argument relation. I have taken the trouble to analyze this text because
I believe it is important for an argumentation analyst to recognize a text
as non-argumentative despite its much resemblance to argumentative texts.
The second text presented here contains characteristics of a narration,
but its real intention is to persuade.
Text 2
01. I come from a middle-class family,
02. and neither of my parents attended college.
03. I also have a younger sister who will be attending college in a couple of years.
04. So when it came down to chose between the Ivy League and affordability,
05. I made a choice that, for me, was obvious - I went to Tulane University,
06. which offered me a full four-year scholarship.
07. Shouldn’t students go to the college that offers the best chance to study,
08. and not to work three jobs,
09. lose sleep,
10. over growing debts
11. and force their family into hardship?
12. I have absolutely no regrets.

13. There are plenty of opportunities available at every school for strong students who
have the ambition, self-confidence and drive to succeed.
14. After all, in four years, when you graduate, what you have achieved will reflect far
more about you than where you achieved it.
Marissa Caudll
New Orleans, LA.
With regard to Text 2, I prefer to concentrate on what I consider to be
argument relations and the place they occupy in the hierarchical structure
of the text, paying less attention on the RST analysis as such.
The text should be split into two parts: 1–6 is only a Background
satellite intended to increase ‘R’s ability to comprehend N’ (Mann and
Thompson, 1988, p. 273), and 7–14 is the nucleus. The nucleus contains
an argument relation: its conclusion/opinion (i.e., N) is presented in 7 +
12 (as it is explained bellow, 7 and 12 present the same opinion: they both
contend, in different ways, that it is preferable for students to go to the
college that offers the best chance to study), and it has six independent
supportive arguments (six Evidence satellites): 8, 9, 10, 11, 13, and 14. It
is important to realize that the apparent question and answer (7–11 and
12, respectively) are not actually a real question/answer (corresponding to
RST Solutionhood relation). The question in unit 7–11 is a rhetorical
question and, in fact, provides four reasons (8 through 11) for the conclu-
sion stated in 7, which really means that students should go to the college
that offers the best chance to study. This same conclusion is also presented,
in another way, in unit 12.
Text 2 is, therefore, a strong and concentrated supportive argumenta-
tive text, in the sense that all the rhetorical relations in the nuclear part of
the text (7–14) are Evidence relations , i.e., supportive argument relations.
The Background relation in the top level of the hierarchical structure serves
only as a precondition to the comprehension of the argumentative part.
Figure 3 below shows this structure very clearly.
110 M. AZAR
Figure 3. A partial RST diagram of Text 2.

One can see, in passing, how the roles of connectives and syntactical
forms are small for rhetorical structure, i.e., the ways in which parts of a
text work together to produce communicative end. Text 2 is a good example
of a linguistic packaging having an appearance of something different than
its real goal.
The next text is also argumentative, but in a complete different way.
Figure 4 presents its rhetorical structure.
Text 3
1. Surely, the removal of the idol as a concrete symbol of religious worship was a
considerable achievement in the age in which it occurred.
2. But this achievement is mistakenly considered as comprehensive,
3. whilst it was only partial.
(translated from a Hebrew newspaper)
Figure 4. RST diagram of Text 3.
The diagram shows that text 3 is an argumentative text, since its two
rhetorical relations are argument relations. The arguments (the satellites)
may be considered as not convincing because no reason at all was given
to support the conclusion (unit 3), but the writer’s intention to use them to
increase the readers’ positive regard of the conclusion renders them per-
suaders. These two persuaders are not completely lacking persuasion force.
In our text, the conclusion does not present the kind of a disputed opinion,
which has its protagonists and its antagonists, and the writer places
him/herself with the former trying to convince the latter or the non-com-
mitted. Rather, the writer came with something new, something the
addressee or readers were not aware of, a new idea, and he/she introduced
that new idea by, first, ‘making a concession’, and acknowledging the
considerable achievement, and, second, by rejection the thesis that con-
tradicts his/her antithesis. In a case like this, the two persuaders seem to
add credibility to the writer.14
To conclude, I would say, first, that the question what is an argumenta-
tive text receives now a precise answer: a text which its nucleus (or nucleus

of nucleus, or nucleus of nucleus of nucleus . . .) is a conclusion and its
satellite (by definition) an argument. Secondly, RST with its five rhetor-
ical relations that have been recognized as argument relations and the four
types of argument that have been distinguished (supportive, incentive,
justifier, persuader) gives effective tools for describing, analyzing and
comparing argumentative texts.
NOTES
1
Moore and Pollack (1992) and Elhadad (1995) point out difficulties when using RST for
text generation and for interactive dialogues, but RST should be considered first of all as a
tool for analyzing monologue written texts.
2
In Mann and Thompson (1988, p. 255), the locus of Concession is the satellite and the
nucleus, but in Mann and Matthiesen (1991), the locus is, like all the other presentational
relations the nucleus alone.
3
Abelen et al. (1993, p. 334) consider Restatement and Summary as two textual relations.
They also view Solutionhood, Interpretation and Evaluation as interpersonal (that is, pre-
sentational).
4
RST was created only for written texts, and, therefore, only writer and reader are used in
RST.
5
See footnote 2. I do not see any reason why Concession should differ from Antithesis in
matter of locus of effect, since both intend to increase positive regard.
6
Barton (1995) sees the two contrastive relations, antithesis and concession, as typical to
academic argumentative discourse. Ducrot (1992, p. 155), says that reading Le Monde is
fascinating from the point of view of the ‘incredible cascade of concessions.’
7
See Azar (1997) for a more detailed discussion of the complexity of concession
relations, and their use as argumentation.
8
One of the advantages of RST over other methods of text analysis is that it allows the
analyst to decide which parts of the text will be broken into its minimal text spans and
which will be broken into larger text spans. The aim of the analysis should determine the
length of the spans, and our aim is to decide if the text, or part of it, is argumentative or
not. One should also remember, when judging a proposed analysis of a text, that there is
often more then one interpretation of a given text. The real important question is: Is the
suggested interpretation a likely one?
9
Wyatt and Badger (1993) convincingly make a distinction between argumentation and
criticism, a distinction usually not found in argumentation literature.
10
The effect of Non-volitional cause is (Mann and Thompson, 1988, p. 275): ‘R recog-
nizes the situation presented in S as a cause of the situation presented in N.’
11
In Azar, 1997, three types of concession relations have been distinguished: a non-
argumentative, a persuader and a supportive.
12
The effect of Non-volitional result is (Mann and Thompson, 1988, p. 276): ‘R recog-
nizes that the situation presented in N could have cause the situation presented in S.’
13
The effect of Volitional cause is (Mann and Thompson, 1988, p. 274): ‘R recognizes
the situation presented in S as a cause for the volitional action presented in N.’
14
Of course, only meticulous empirical researches can provide answers concerning the
usage of concession and antithesis relations in argumentative texts.
112 M. AZAR

REFERENCES
Abelen, E., G. Redeker and S. A. Thompson: 1993, ‘The Rhetorical Structure of U.S.-
American and Dutch Fund-Raising Letters’, TEXT 13, 323–350.
Azar, M.: 1995, ‘Argumentative texts in newspapers’, in F. H. van Eemeren, R. Grootendorst,
J. A. Blair and C. H. Willard (eds.), Proceedings of the Third ISSU Conference on
Argumentation, University of Amsterdam, June 21–24, SicSat, Amsterdam, vol. 3, pp.
493–500.
Azar, M.: 1997. ‘Concession Relations as Argumentation’, TEXT 17, 301–316.
Barton, E. L.: 1995, ‘Contrastive and Non-contrastive Connectives Metadiscourse Functions
in Argumentation’, Written Communication 12, 219–239.
Blair, A. J.: 1992, ‘Everyday Argumentation from an Informal Logic Perspective’, in W. L.
Benoit, D. Hample and P. J. Benoit (eds.), Readings in Argumentation, Foris, Berlin-New
York, pp. 357–378.
Brassart, D. G.: 1996a, ‘Didactique de l’argumentation écrite: Approches psycho-cognitives’,
Argumentation 10, 69–87.
Brassart, D. G.: 1996b, ‘Does a Prototypical Argumentative Schema Exist? Text Recall in
8 to 13 Years Olds’, Argumentation 10, 163–174.
Brockriede, W.: 1990, ‘Where is Argument?’, in R. Trapp and J. Schuetz (eds.), Perspectives
on Argumentation, Essays in Honor of Wayne Brockriede, Waveland Press, Prospect
Heights, pp. 4–8.
Chittleborough, P. and M. E. Newman: 1993, ‘Defining the Term “Argument”’, Informal
Logic 15, 189–207.
Connor, U and J. Lauer: 1985, ‘Understanding Persuasive Essay Writing: Linguistic/
Rhetorical Approach’, TEXT 5, 309–326.
Crosswhite, J.: 1996, The Rhetoric of Reason: Writing and the Attractions of Argument,
The University of Wisconsin Press, Madison.
Ducrot, O.: 1992, ‘Argumentation et persuasion’, in W. De Mulder, F. Schuerewegen and
L. Tasmowski (eds.), Enonciation et parti pris, Actes du colloque de l’Université d’Anvers
(5, 6, 7 Fevrier 1990), Rodopi B.V., Amsterdam, pp. 143–158.
van Eemeren, F. H., R. Grootendorst and T. Kruiger: 1987, Handbook of Argumentation
Theory: A Critical Survey of Classical Backgrounds and Modern Studies, Foris,
Dordrecht.
Elhadad, M.: 1995, ‘Using Argumentation in Text Generation’, Journal of Pragmatics 24,
189–220.
Fox, B.: 1987, Discourse Structure and Anaphora in Written and Conversational English,
Cambridge University Press, Cambridge.
Golder, C. and P. Coirier: 1996, ‘The Production and Recognition of Typological
Argumentative Text Markers’, Argumentation, 10, 271–282.
Grize, J.-B.: 1981, ‘L’argumentation: explication ou séduction’, in L‘argumentation, Presses
universitaires de Lyon, Lyon, pp. 29–40.
Hovy, E. E.: 1991, ‘Approaches to the Planning of Coherent Text’, in C. Paris and W.
Swartout (eds.), Natural Language Generation in Artificial Intelligence and
Computational Linguistics, Kluwer Academic Publishers, pp. 83–102.
Knott, A.: 1991, New Strategies and Constraints in RST-Based Text Planning, Ms.C. Thesis,
Department of Artificial Intelligence, University of Edinburgh.
Linden, K., S. Cumming and J. Martin: 1992, ‘Using System Networks to Build Rhetorical
Structures’, in Proceedings, Sixth International Workshop on Natural Language
Generation, Heidelberg, Berlin.
Mann, W. C. and C. M. I. M. Matthiessen: 1991, ‘Functions of Language in Two Frame-
works’, Word 42, 231–250.
Mann, W. C., C. M. I. M. Matthiessen and S. A. Thompson: 1992, ‘Rhetorical Structure
Theory and Text Analysis’, in W. C. Mann and S. A. Thompson (eds.), Discourse

Description: Diverse Linguistic Analyses of a Fun-Raising Text, Benjamins, Amsterdam,
pp. 39–78.
Mann, W. C. and A. Thompson: 1986, ‘Relational Propositions in Discourse’, Discourse
Processes 9, 57–90.
Mann, W. and S. Thompson: 1987, ‘Rhetorical Structure Theory: Description and
Construction of Text Structure’, in G. Kempen (ed.), Natural Language Generation,
Martinus Nijhoff, The Hague, pp. 85–96.
Mann, W. C. and S. A. Thompson: 1988, ‘Rhetorical Structure Theory: Toward a Functional
Theory of Text Organization’, TEXT 8, 243–281.
Martin, J. R.: 1992, English Text – System and Structure, John Benjamin Publishing Co.,
Philadelphia-Amsterdam.
Matthiessen, C. and S. A. Thompson: 1988. ‘The Structure of Discourse and Subordination’,
in J. Haiman and S. A. Thompson (eds.), Clause Combing in Grammar and Discourse,
Benjamins, Amsterdam, pp. 275–330.
Moore, J. D. and C. L. Paris: 1993, ‘Planning Text for Advisory Dialogues: Capturing
Intentional and Rhetorical Information’, Computational Linguistics 19, 651–694.
Moore, J. D. and M. E. Pollack: 1992, ‘A Problem for RST: The Need for Multi-Level
Discourse Analysis’, Computational Linguistics 18, 537–544.
O'Brient, T.: 1995, ‘Rhetorical Structure Analysis and the Case of Inaccurate, Incoherent
Source-Hopper’, Applied Linguistics 16, 442–482.
Perelman, C.: 1982, The Realm of Rhetoric, University of Notre Dame, Notre Dame.
Robrieux, J-J.: 1993, Eléments de rhétorique et d’argumentation, Dunod, Paris.
Rosner, D. and M. Stede: 1992, ‘Customizing RST for the Automatic Production of Technical
Manuals’, in Proceedings, Sixth International Workshop on Natural Language
Geneneration, Heidelberg, Berlin, pp. 199–215.
Thompson, S. A. and W. C. Mann: 1987, ‘Antithesis: A Study in Clause Combining and
Discourse Structure’, in R. Steele and T. Threadgold (eds.), Language Topics, Essays in
Honour of Michael Halliday, John Benjamins B.V., Amsterdam, vol. 2, pp. 359–379.
Toulmin, Stephen: 1969, The Use of Argument, Cambridge University Press [1st edn. 1958].
Wyatt, R. O. and D. P. Badger: 1993, ‘A New Typology for Journalism and Mass
Communication Writing’, Journalism Educator 48, 3-1.
114 M. AZAR

Argumentative text as rhetorical structure An application of rhetorical structure theory.pdf

More Related Content

Similar to Argumentative text as rhetorical structure An application of rhetorical structure theory.pdf (20)

More from Cynthia Velynne (20)

Recently uploaded (20)

Argumentative text as rhetorical structure An application of rhetorical structure theory.pdf