These are analyses that either were done during the development of
RST or more recently, by the developers. They have all been reviewed
for consistency and currency.
The right hand column shows counts of the signaled relations of each
of these texts, along with the total relation count. Signals here include
conjunctions, phrases, open class words and syntactic patterns. Methods
of counting signals have not been formalized, and there may not be
a need to do so.
This is not a balanced collection in any sense. There was a tendency to seek diversity and wide coverage of genres rather than balance, and after that some texts have been set aside so that they can be presented in topical collections. It does not represent spontaneously produced texts, and none of the texts are drawn from dialogues.
Some of the texts have been chosen because they illustrate some problem, limitation or weakness of RST. This is intended as an encouragement to other researchers to develop accounts in the indicated directions.
Two Frameworks: RST and SFL -- PDF Two Frameworks: RST and SFL -- GIF
This text is the title and abstract of the article indicated. Like the
other abstracts on the website, it is a complete text in the sense that
abstracts are in part intended to allow the reader to decide whether to
read the article, and for that purpose cannot rely on the unread text of
The text was also chosen because it describes the relationship between
Systemic Functional Linguistics and RST.
We are very interested in developing knowledge of how RST compares with
other approaches to discourse analysis. So, we expect to add to the site
some analyses or links to alternate analyses of the same or similar texts.
Reference: WORD, 42 (3), William C. Mann and Christian M. I. M. Matthiessen,
December 1991, p. 231-249.
Lactose and Lactase -- PDF Lactose and Lactase -- GIF
This abstract is nearly half background information. One of the reasons
that backgrounding is not signalled by its own distinct signal (such as
a conjunction) may be that such a signal would in effect say that the author
thought the reader would not know this information. Some readers might
find that mildly condescending if it were made explicit.
Notice that one of the key relationships in the text is an analogical relationship
between "Smile at your husbands" and "I find it hard sometimes
to smile at Jesus." Notice also that there is no trace of this analogical
relationship in the RST analyses of the text. It is certainly a discourse
structural element, and it is certainly involved in the way that the text
communicates, but it is of a different order than RST identifies.
There are inherent limits on how much a monologue oriented method such
as RST can find in text containing dialogue. One of the large challenges
raised by CRST is to provide some comparable structured whole-text or whole-event
representation, first for natural two-party dialogue and eventually for
It is also notable for the depth of the tree, 16 levels (including the
whole text level), and for the elaborate background portion. Is the strategy
of using an elaborate background portion related to opposing a respected
Sparky Lived! -- PDF -- coherence and imagination Sparky Lived! -- GIF
This little text is the main body of an advertisment for V8 vegetable juice.
It is notable for the extreme feat of imagination that it requires in order
to see it as coherent. It is also notable for the lack of cohesive signals,
especially that there is no clear signal of cause.
The text is addressed to "you." So you must imagine, possibly
against personal reality, that:
1. As a child, you had a pet;
2. the pet was named "Sparky;"
3. you threw vegetables under the table (presumably to avoid eating them);
4. Sparky ate them;
5. Eating the vegetables affected Sparky's health.
The reader's search for coherence leads to a kind of assertion, in RST
terms a Relational Proposition.
The notion that the vegetables caused Sparky to live a long time relies
on a generalization that eating vegetables tends to cause longevity. It
is implicit and defeasible (deniable). It would probably be illegal to
make it explicit, because that would be a medical claim, and medical claims
are regulated by law. Legal aspects are probably enough of a threat that
they are the likely reason for including the fine print disavowal of the
claim (see second image.)
The text also illustrates how a background photo can function in a preparatory
way, just as the title of a paper does. Without the photo as a leading
preparation, the search for coherence might tend to fail.
It is interesting to consider how this sort of text is related to fiction.
Certainly, for the advertiser's purposes, the assertion of cause does not
belong entirely to the imagined world created by the reader. It must also
affect the world in which the reader buys groceries. So, this is more like
a parable than it is like a fictional drama.
Copyright Notice -- PDF -- about not communicating Copyright Notice -- GIF
The significance of this text for RST is that it is not really a
communicative text, so the structure is filled with the Joint relation, identifying
unrelated contributions upward.
There is only a very local topical sort of organization. The gross structure
is a list of terms. In terms of intentions, the author intended to create
and limit rights of use relative to this intellectual property. This
has nothing in common with changing the reader, and so is not communicative.
This failure to effectively analyze noncommunicative text illustrates
that RST has responded to some essential features of communication.
Devotional Reading -- PDF -- issues of linearity Devotional Reading -- GIF
One interest in this devotional page is that it shows an example where
effective understanding of the text seems to require recognizing an event
in the story which is prominent, but not represented by any text. In
the analysis diagram it is represented by "GHOST SEGMENT," along
with imputed content. Ellipsed segments of this sort are rare in written
Another interest is that there is an important part of the organization
of this text that is carried by a strong template of layout and formatting.
Fonts, whitespace, break lines and positions are employed to give simultaneous
approachability to several parts. (The analysis is confined to the body
of the right hand page.)
RST is implicitly working with a linear view of text and so does not
have a representation for these features.