Rhetorical Structure Theory and Text Analysis

Abstract

Recent research on text generation has shown that there is a need for stronger linguistic theories that tell in detail how texts communicate. The prevailing theories are very difficult to compare, and it is also very difficult to see how they might be combined into stronger theories. To make comparison and combination a bit more approachable, we (Mann and Thompson) have created a book which is designed to encourage comparison. A dozen different authors or teams, all experienced in discourse research, are given exactly the same text to analyze. The text is an appeal for money by a lobbying organization in Washington, D.D. It informs, stimulates and manipulates the reader in a fascinating way. The joint analysis is far more insightful than any one team's analysis alone. This paper is our contribution to the book. Rhetorical Structure Theory (RST), the focus of this paper, is a way to account for the functional potential of text, its capacity to achieve the purposes of speakers and produce effects in hearers. It also shows a way to distinguish coherent texts from incoherent ones, and identifies consequences of text structure. Keywords: Artificial intelligence; Coherence; Computational linguistics; Discourse analysis; Linguistics; Natural language; Rhetorical structure theory; Text structure.

Full text

ISI
Research
Report
ISIIRR-89-242
FILE
COPY
November
1989
AD-A222
655
W
illia
m
C
.M
a
n
n
0
cf
,'
n
i
,a -
',
Christian
M. I. M.
Matthiessen
Sandra
A.
Thompson
..........
Rhetorical
Structure
Theory
and
Text
Analysis
DTIC
ELECTE
-DEMI3,I:
ON
STA
-NM
1
Appmoyd
for
Puc
Mwk aq
INFORMATION
SCIENCES
21TA;
,22-,I -
INSTITUTE!w24,6Amrii
L66Amrt'i iAaiaJ.
c/a:~~a~'~~~'

Unclassified
SECURITY
CLASSIFICATION
OF THIS PAGE
REPORT
DOCUMENTATION
PAGE
la.
REPORT SECURITY
CLASSIFICATION
lb.
RESTRICTIVE
MARKINGS
Unclassified
2a.
SECURITY
CLASSIFICATION
AUTHORITY
3.
DISTRIBUTION
/AVAILABILITY
OF
REPORT
2b.
DECLASSIFICATION
I
DOWNGRADING
SCHEDULE
This
,ment
is
approved
for
public
release;
distribution
is
unlimited.
4.
PERFORMING
ORGANIZATION
REPORT
NUMBER(S)
S.
MONITORING ORGANIZATION
REPORT
NUMBER(S)
ISI/RR-89-242
6a.
NAME
OF
PERFORMING
ORGANIZATION
6b.
OFFICE
SYMBOL
7a.
NAME
OF
MONITORING ORGANIZATION
USC/Information
Sciences
Institute
(if
appi
able)
6c.
ADDRESS
(City,
State, and
ZIP
Code)
7b.
ADDRESS
(City,
State,
and
ZIP
Code)
4676 Admiralty
Way
Marina
del
Rey,
CA
90292-6695
8a.
NAME
OF
FUNDING/SPONSORING
l8b.
OFFICE
SYMBOL
9.
PROCUREMENT INSTRUMENT IDENTIFICATION
NUMBER
ORGANIZATION
Of
a
cabe)
--
over--
NSF,
AFOSR,
DARPA
Sc.
ADDRESS(City,
State,
and
ZIP Code)
10.
SOURCE
OF
FUNDING
NUMBERS
PROGRAM
PROJECT
TASK
WORK
UNIT
--
over--
ELEMENT
NO.
NO.
NO.
ACCESSION
NO.
11.
TITLE
(include
Security
Clasification)
Rhetorical
Structure
Theory
and
Text
Analysis
(Unclassified)
12.
PERSONAL
AUTHOR(S)
Mann,
William
C.;
Matthiessen,
Christian
M.
1.
M.;
Thompson, Sandra
A.
13a.
TYPE
OF
REPORT[113b"
TIME
COVERED
114" DATE
OF
REPORT
(Year,Moth'
ay)
is'
PAGE
COUNT
67
3b
FRO
14____T
1989,
November
AG
OUT
6
16.
SUPPLEMENTARY
NOTATION
17.
COSATI
CODES
18.
SUBJECT
TERMS
(Continue
On
reverse
if
necessy
and
identify
by
block
number)
FIELD
GROUP
SUB-GROUP
"artificial
intelligence,
coherence,
computational
linguistics,
discourse
analysis,
09
02
linguistics,
natural
language,
Rhetorical
Structure
Theory,
text
structure',
19.
ABSTRACT
(ContInue
on
reverse
if
necessary
and
identify
by
block
number)
Recent research on
text
generation
has
shown
that
there
is
a
need
for
stronger
linguistic
thoeries
that
tell
in
detail
how
texts
communicate. The
prevailing
theories
are
very
difficult
to
compare,
and
it
is
also
very
difficult
to
see
how
they might
be
combined
into
stronger
theories.
To
make
comparison
and
combination
a
bit
more
approachable,
we
(Mann
and
Thompson)
have
created
a
book
which
is
designed
to
encourage
comparison.
A
dozen
different
authors
or
teams,
all
experienced
in
discourse research,
are
given
exactly
the
same
text
to
analyze.
The
text
is
an
appeal
for
money
by
a
lobbying
organization
in
Washington,
D.C.
It
informs,
stimulates
and
manipulates the reader
in
a
fascinating
way.
The
joint
analysis
is
far
more
insightful
than
any
one
team's
analysis
alone.
This
paper
is
our
contribution
to
the
book.
Rhetorical Structure
fheory
(RST),
the
focus
of
this
paper,
is
a
way
to
account
for
the
functional
potential
of
text,
its
capacity
to
achieve
the
purposes
of
speakers
and
produce
effects
in
hearers.
It
also
shows
a
way
to
distinguish
coherent
texts
from
incoherent
ones,
and
identifies
consequences
of
text
structure.
20.
DISTRIBUTION/AVAILABILITY
OF ABSTRACT
21.
ABSTRACT
SECURITY
CLASSIFICATION
MI
UNCLASSIFIEDIUNLIMITED
0
SAME
AS
RPT.
C3
OTIC
USERS
Unclassified
22a.
NAME
OF
RESPONSIBLE
INDVIDUAL 22b.
TELEPHONE
(Indulde
Area
Code)
22c.
OFFICE
SYMBOL
Victor
Brown
Sheila
Coyazo
213/822-1511
OD FORM
1473,
84
MAR
83
APR
edition
may
be
used
until
exhausted.
SECURITY
CLASSIFICATION
OF
THIS
PAGE
All
other
editions
are
obsolete.
Unclassified

Unclassified
GoUuTVy
C
_AIPCAIOW
Or
TWOs
PAIK
8c.
(continued)
National
Science
Foundation
1800
G
Street
NW
Washington,
DC
20550
Air
Force
Office
of
Scientific
Research
Boiling
Air
Force
Base,
Building
410
Washington,
DC
20332
Defense
Advanced
Research Projects
Agency
1400
Wilson
Boulevard
Arlington,
VA 22209
9.
(continued)
NSF:
IST-8408726
AFOSR:
FQ8671-84-01007,
F49620-87-C-0005
DARPA:
MDA903-87-C-0641
Accession
For
NTIS
GRA&I
DTIC
TAB
::
.Unannounced
Q3
Justification
By
Distribut
ion/
Availability
Codes
Avail
and/or
Dist
Speoial
Unclassified
ISCURITy
CLASUIPICArIOW
OF
THIS
PAGE

ISI
Research
Report
ISI/RR-89-242
November
1989
Universit
William
C.
Mann
California
Christian
M. I. M.
Matthiessen
Sandra
A.
Thompson
*
:
*:*.*.*.*.::*.*.::.:*....
Rhetorical
Structure
Theory
and
Text
Analysis
INFORMATION
SCIENCES
2131822-1511
INSTITUTE
4676
Admiral'
Way/Marina
del
Rey/California
00292-6605
This
research
Is
sponsored
In
part
by
the
National
Science Foundation under
Grant
r,,,,,ber
IST-8408726,
In
part
by
the
Air
Force
Office
of
Scientific
Research
under
Contract
Numbers
FQ8671-84-01007
and
F49620-87-C-0005.
and
In
part
by
the
Defense
Advanced Research
Projects
Agency
under
Contract
No.
MDA903-87-C-0641.
Views
and
conclusions contained
In
this
report
are
those
of
the
authors
and
should
not
be
Interpreted
as
representing
the
official
opinion
of
NSF.
AFOSR.
DARPA,
the
U.S.
Government, or
any
person or agency
connected
with
them.

Table
of
Contents
1
Aims
of
Discourse
Analysis
1
2
Varieties
of
Structure
:3
3
Introduction
to Rhetorical
Structure
Theory
5
3.1
Underlying
Assumptions
6
3.2
Terminology
and
mechanisms
of
RST
11
3.3
Steps
in
performing
an
RST
analysis
16
4
RST
Analysis
of
the
ZPG
Letter
19
5
Alternative
Analyses
32
6
Consequences
and
Inferences from
the
Analysis:
Rhetorical
Patterns
:35
6.1
Centrality
at the
Whole-Text
Level
35
6.2 A
Presentational
Pattern
36
7
Consequences
and
Inferences from
the
Analysis:
Relations,
Grammar
and
37
Lexis
7.1
Relations,
Relational
Propositions
and
Explicit
Signals
37
7.2
"Subordination"
42
7.3
Concessives
44
8
Status
of
RST
45
9
Conclusion
46
I.
Appendix:
Relation
Definitions
48

List
of
Figures
Figure
1:
Definition
of
the
Evidence
Relation
12
Figure
2:
RST
Diagram
of
the
Tax
Program
Letter
13
Figure
3:
RST Diagram
Showing
the
Circumstance
Relation
15
Figure
4:
Highest
Level
RST
Schema
Diagram
for
the
ZPG
Letter
17
Figure
5:
The
Original
Form
of
the
ZPG
letter
21
Figure
6:
The
Segmented Form
of
the
ZPG
letter
23
Figure
7:
Holistic
Structure
Sequence for
the
ZPG
Letter
24
Figure
8:
RST
Diagram
for
the
Body
of
the
ZPG
Letter
25
Figure
9:
The
Relations
of
the
ZPG
Letter
33
Figure
10:
An
Alternate
Analysis of
Units
11
through
16
35
Figure
11:
Satellite-Nucleus-Satellite
Patterns
in
the
ZPG
Text
37

1
Aims of
Discourse Analysis
from
Advanced
Learner's
Dictionary
(Hornby
et
al):
language
n.
1.
humani
and
non-instinctive
method
of
communicating
ideas,
feelings,
and
desires
by
means
of
a
system of
sound
symbols.
2....
The
key
distinguishing
attribute
in
defining
"language"
is
that
it
is
a
metaod
of
communicating.
When
people
get
together and
interact,
especially
in
frequently
recurring kinds
of
encounters,
our
usual
experience
is
that
the
interaction
is
purposeful
for
one
or
more
participants.
They
can
achieve
particular
ends
by using
language.
If
speaker
and
hearer
are
using
a
shared
language
then
frequently they
can achieve
their
purposes,
but
if
they
do
not
have
a
shared
language
then
they
fail.
Communication
is
effective
if
it
creates
the
potential
for
a
successful
outcome,
achievement
of
the
participants'
purposes.
Invitations
to
social
events,
negotiations
to
buy
things
and
requests
for
gifts
of
money
are
a
few
of
these
kinds
of
encounters.
All
this
would
be
tediously
obvious,
except
that
this
capacity
of
languages
to
enable
success
in
such
interactions
has not
been
accounted
for by
linguistics.
In
our
research
in
discourse
analysis,
we
aim
to
contribute
to
an
account
of
how
language
can
be
used
to
communicate,
i.e.
a
detailed description
of
how
it
contributes
to
the
outcomes
of
interactions.

2
Pleading
for
money has
a
particularly
long social
history.
It
is
a most
representative
class
of language
use
encounters. The
form
we
study
here
(a
letter
to
a
large
set
of
people
from
a
corporation
organized
for
a
narrow
social
cause)
is
contemporary.
An
essential
part
of
success
in
this
encounter
is
of
course
having
a
suitable number
of
addressees
give
money.
To
create
the
potential
for
such
an
outcome.
the
addressees
must:
*
know
that
money has
been
requested,
*
know
how
to
comply
with
the
request,
*
be
motivated
and
willing
to
respond.
So
to account
for
the
ZPG
letter
as
communication
requires
that
we
account
for
how
it
can
produce
at
least these
effects.
To
use
language
as
in
the
ZPG
letter
is
not
merely
performing
in
some
role
or
ritual
as
an
institutional
beggar;
playing
out
a role
can
be
done
as
an
activity
without
consequences,
and
a
view
of
language
simply
in
terms
of
social
roles
would
miss
the
point.
Rather,
the
attempt
to
induce
people
to
choose
to
give
money
is
a
characteristic.
if
not
defining,
attribute
of
this
class
of
encounters.
It
is
the
attempt
to get
money
by
symbolic
means
that
is
the
communicative
essence
of
this
class
of
encounters.
More
generally,
we
want
our
discourse
theories
to
provide
paths
or
mappings
both
from
situation
to
language,
explaining
how
or
why
particular
uses
of
language
were
chosen,
and
from
language
and
situation
to effect,
explaining
why
particular
uses
of
language
succeeded
or
failed.
For
the
latter,
the
most
obvious
fulfillment
in
the theory
presented
here
is
that
it
imputes
to
the
speaker
desires
for
particular
effects. This
element
distinguishes
it
from
many
other
approaches
to
discourse
description.
Most

3
often,
discourse
descriptive
methods
have
no
place
for
intended
effects.
although
frequently
the
developers
of
the
methods
will
acknowledge
them.
Certain
kinds
of
presentations,
texts,
have
a
kind
of wholeness
or
integrity
that
others
lack.
We
recognize
that
they
"hang
together"
and
are
understandable
as
single
objects.
They
are
coherent. Every
element
has
some
role in
the
whole
text:
otherwise
the text
contains
a
nonsequitur.
This
is
the
sense
in which
we
see
magazine
articles
as
texts,
but
magazines,
news
broadcasts,
and
some
dialogues
as
structured
collections
of
texts.
A
theory
of
textual
communication
should
account
for
this
coherence:
if it
is
an
organizational
theory
then
it
should
account
for
the
presence
or absence of
nonsequiturs.
To
account
for
nonsequiturs,
the
theory
must
assign
(or
appropriately
fail
to
assign)
a
status
to
every
part
of
the
text.
One
of
our
aims
in
creating
a
theory
of
communication
is
this
kind
of comprehensive
assignment
of
status.
This
is
not
to
say
that
there
is
anything
wrong
with
selective
commentary
or
partial
theories,
except
that
they
cannot
by
their
nature
account
for
the
impression
of
unity
or
integrity
that
is
one
of
the
identifying
marks
of
texthood.
2
Varieties
of
Structure
The
term
"structure"
in
this
paper
is
used
in
an
organizational
sense.'
The
name of
the
theory, Rhetorical
Structure
Theory
(RST),
employs
this
sense.
A
theory
of
text
structure
describes
what
sorts
of
parts
texts
have,
and
the
principles
of
combination
of
parts into
entire texts.
Since
the
term
is
widely
used
in
many
senses,
we
IThis
is
distinct
from some
other
uses
in
discourse
analysis,
e.g.
for
describing
referential
recurrence.

4
attempt
to
sketch
here
the
various kinds
of
structure
that
RST
recognizes,
and
within
those
the
scope
of
the
theory itself.
Even
in
our
narrow
sense
of
the
term, RST
does
not
attempt
to describe
all
of
the
kinds
of
structure
that
a
text
may
have.
It
recognizes
three
principal
kinds:
1.
Holistic
Structure
--
structure
deriving
from
the properties
of
the
genre
or
variety
of
text,
used in
this
case
to
describe
the
form
of
letters
and
account
for
expressions
such
as
Sincerely.
2.
Relational
Structure
--
structure
expressing
the organization
of
coherent
contiguous
text,
used
in
this
case
to
describe
the
internal
composition
of
the
body
and
P.S.
of
the
letter.
3.
Syntactic
Structure
--
as
the
term
is
commonly
used.
2
As
will be
seen
below,
RST
is
a
theory
of
relational
structure.
It
acknowledges
other
kinds,
interacting
strongly with
holistic
and
syntactic
structure,
but
it
does
not
attempt
to
incorporate
accounts
of
either
genre
or
syntax.
In
the
interaction,
there are
alignments of
the various
sorts
of
structure,
so
that
often
a
particular
arrangement,
such
as
a
combination
of
clauses,
can
be
described
in
more
than
one
way.
The
varieties
of
structure
interpenetrate
and illumine
each
other,
making
descriptions
in
terms
of
these
perspectives
particularly
interesting.
One of
the
consequences of
this
interpenetration
is
that
there
is
no
sharp
boundary, either
in
principle
or
in
practice,
2
RST
can
interface with
different
theories
of
syntacti:
structure.
However,
the
nature
of
the 'interface'
between
rhetorical
structure
and
syntactic
structure
may
vary
quite
significantly
depending
on
the
nature
of
the
syntactic
theory.
In
general,
functional
theories
of
syntactic
structure
will
simplify
the interface
considerably.
The
treatment
of
so-called
subordinate
clauses
is
a
case
in
point:
see
Section
7.2.

5
between
the
different
varieties.
3
3
Introduction to
Rhetorical
Structure
Theory
For
several
years
there
has
been
a
research
effort
at
USC
Information
Sciences
Institute
(ISI)
aimed
at
describing
written
discourse.
The
original
motivation
for
this
research
was
that
its
results
might
be
useful
as
a
theoretical
basis
in
designing
computer
programs
with
some of
the
capabilities
of
authors.
The
effort
has
ranged
beyond
this
limited
goal,
and
resulted
in
an
understanding
of
discourse
that
has had
many
other
uses,
including
several
applications
in
linguistics.
The
effort
has
involved
a
study
of
the
nature
of
text
as
a
medium
of
communication,
with
an
interest
in
developing
a
theory
of
text
structure
that
could serve
both
as
an
analytical
tool
and
as
a
tool for
text
generation.
We
call
this
theory Rhetorical
Structure
Theory (RST),
since
it
provides
a
framework
for
describing
rhetorical relations
among
parts
of
a
text.
4
In
the
construction
of
this
theory
we
have
analyzed
more
than
400
texts,
from
one
paragraph
to
several
pages
in
length,
of
the
following
types:
administrative
memos.
3
1n
RST,
the
approach to theorizing
is
to
treat
theories
as
components
of
an
account
rather
than
as
monolithic.
While
we
do
not
select
any
particular
theory
of
holistic
structure,
the
systemic
theory
of
generic
structure,
[Hasan
781,
[Halliday
&
Hasan
88],
or
macrostructure
theory,
[van
Dijk
72],
[van
Dijk
77],
[van
Dijk
801
offer
possibilities.
In
each
case
the
theory
provides
much more
than
just
a
correlate
of
holistic
structure.
RST
also
has
strong interactions
with
independent
theories
of
Thematic
Structure.
i.e.
the
establishment
and
maintenance
of
topics,
and with
Exchange
Structure,
i.e.
dialogue
structure,
but
these
interactions
are beyond
the
scope
of
the
paper
(see
Section
8).
4
We
gratefully
acknowledge
the
valuable
input
provided
by
Cecilia
Ford.
Barbara
Fox.
and
Peter
Fries
in
the
development
of
RST.
We expres3
special
thanks
to
Peter
Fries
for his
comments
on
the
analysis
of
the
ZPG
text.
RST extends
a
tradition
of
research
on
the relational
basis
of
text
structure.
The unabridged
version
of
[Mann
&
Thompson
89]
(
[Mann
&
Thompson
87a])
contains
a
substantial
discussion
of
the
relationships
between,
and
the
influence
upon,
RST
and
other
relational studies
of
text,
including
the
work
of
Beekman,
Callow,
Grimes.
Grosz,
Halliday,
Hasan,
Hobbs.
Hoey,
Jordan,
Kopesec.
Longacre,
Martin.
Mckeown,
Meyer,
Pike.
Sidner
and
Winter. Other
major
papers
on
RST
include
[Mann
&
Thompson
87bl
and
[Mann
&
Thompson
881.

6
personal
letters,
letters
to
the
editor, advertisements,
Scientific
American articles
and
abstracts,
newspaper
articles and
editorials, organizational newsletter
articles
and
appeals
(as
exemplified
in
the
Zero
Population
Growth
letter
that
is
the
focal
text
of
this
book),
public
notices
in
magazines,
travel brochures,
and
recipes.
In
the
course
of
examining
these
texts.
we
observed
that
many
phenomena
of
text
structure
involved
pairs of
regions
of
the
text.
The
mutual
relevance
of
the
two
parts,
and sometimes
their
position
and
form,
could
be
identified
with
recurrent
relations
holding
between the
parts.
These
relations,
sometimes
but
not
always
indicated
by
conjunctions,
could
hold
between
text
parts
of
a
wide
range
of
sizes.
from
clauses
to
groups
of
paragraphs.
These
observaions
led
to the
formation
of
a
testable
set
of
assumptions
(described
below)
and
to
the
realization
of
these
assumptions
in
the
mechanisms
of
RST.
RST
describes
texts
in
a
rich
and
highly
constrained
way
-nd
thus
predicts
much
about their character
and
effects.
It
describes
functions and
structures
that
make
texts
effective
and comprehensible
tools
for
human
communication.
3.1
Underlying
Assumptions
Our
observations
about
text
structure
have
led
to
a
number
of
basic
assumptions
underlying
RST:
1.
Organization
--
Texts
consist
of
functionally significant parts:
the
parts
are
elements
of
patterns
in
which
parts
are
combined
to
create
larger
parts
and
whole
texts.
The
assumption
that
text
is
organized
is
not
controversial:
the
opposite
--
that
texts
do
not
commonly
have an
internal
organization
-- is
not
defended
seriously
in
the linguistic
literature.
2.
17nty
and
Coherence
--
To
be
recognized
as
a
text,
the writing
must create
a
sense
of overall
unity
to
which every
part
contributes.
The
presence of
this
unity
and
coherence
is
uncontroversial,
but
there
are
diverse
views
of
its
sO
iirce.

7
3.
Unity
and
Coherence
Arise
From
Imputed
Function
--
A
(region
of)
text
is
perceived
as
having
unity
and
coherence because
all
of
its
parts
are
seen
as
contributing
to
a
single
purpose
of
the
writer,
i.e.
as
created
to
achieve
a
single
effect.
As
an
alternative,
some
have
assumed
that
unity
and
coherence
come
from
conformity
to
a
familiar
pattern
in
the subject
matter.
such
as
a
temporal
sequence
or
repeated
reference
to
a
character. Others
find
it
in
more
abstract
se(mantic
patterns,
such
as
hyponomy
and
metonymy.
4.
Hierarchy
--
Text
are
organized such
that
elementary
parts
are composed
into
larger
parts,
which
in
turn
are
composed
into
yet
larger
parts
up
to
the
scale
of
the
whole
text.
Without
specifying
the
nature
of
the
parts
or
the
principles
of
composition,
the
assumption
of
hierarchy
contrasts
with
other
assumptions
about
the
patterns
of
text
structure.
For
example,
one
could
assume
that
text
structure
is
formed
by
adjacency
patterns
or
by
linearly
related
chains
of
clauses or
semantic propositions.
5.
Homogeneity
of
Hierarchy
--
As
indicated
above,
RST
describes
relational
structure
and
its
interaction
with
holistic
and
syntactic
structure.
Within
relational
structure,
RST
assumes
homogeneity:
there
is
one
set
of
structural
patterns
available
for
organizing
the
text
at
every
scale,
from
the
largest,
an
element
of
holistic
structure
(e.g.
letter
body, magazine
article
body,
possibly
the
whole
text...)
down
to
the
smallest
scale
(possibly a
two-clause
combination). This
set
of
patterns
is
identified
as
RST
schemas
(to
be
further
discussed
below).
The
potential
for
relational organization
does
not
vary
with
scale;
frequencies
will
vary
with
scale,
genre
and
other
influences.
There
are
no
conventional
patterns
at
scales
between
the
RST
schema
and
the
element
of
holistic
structure.
An
alternative
assumption
might
be
that
there
is
a
rank-scale
or
size-scale
of
objects,
e.g.
sections and
paragraphs.
which
each
have
their
own
distinct
functional
descriptions
and principles
of
relational
composition.
6.
Relational
Composition
--
The
principal
structural pattern
in
multisentential
text
is
relational:
a
small
set
of
highly
recurrent
relations
holding
between
pairs
of
parts
is
used
to
link
parts together
to
form
larger
parts.
There are
several
kinds
of
alternative assumptions
used
by
various researchers.
In
one.
structural
patterns
are
patterns
of
constituent
categories
(analogous
to the
mechanisms
of
certain
grammars).
In
another,
structural patterns
are
by
nature
semantic:
they
are necessarily
patterns
of
subject
matter,
e.g.
temporal
or
causal chains.
Note
that
RST
does
not
assume
that
all
structuring
is
relational,
nor
that
relational
structure
excludes
semantic
structuring,
nor
that
all
patterns
are
based
on
simple
pairs.
The
RST
assumption
is
that
relational
patterns
are
strongly
dominant.
7.
Asymmetry
of
Relations
--
The
most
common
type
of
text
structuring
relation
is
an
asymmetric
class,
called
nucleus-satellite relations
in
RST.

8
This
class
is
asymmetric
because
one
member
of
a
pair
of
text
spans
is
more
central
(the
nucleus)
and
one
more
peripheral
(the
satellite).
Further,
a
text
part
that
is
the
nucleus
for
some
text-structuring
relation
will
have
functional
similarities
with
other
nuclei.
There
are
other
theories
of
text
structure
that
also
recognize
this asymmetry
(cf.
[Grimes
75],
[van
Dijk
85],
Ivan
Dijk
81]
and
the
Longacre.
Meyer
and
Pike
papers
in
the
volume
that
is
to
include
this
paper,
[Mann
&
Thompson
90],
as
well
as
the
references
that
they
cite.)
8.
Nature
of
Relations
--
Text
structuring
relations
are
functional;
the
character
that
they
all
share
can
be
stated
in
terms
of
the
categories
of
effects
that
they produce.
They
can
be
described
in
terms
of
the purposes
of
the
writer,
the
writer's
assumptions
about
the
reader,
and
certain
propositional
patterns
in
the subject
matter
of
the
text.
The
text
structuring
relations
reflect
the
writer's
options
of
organization
and
presentation;
it
is
in
this
sense
that
an
RST
structure
is
"rhetorical."
T
in
contrast,
one
could
assume
that
text
structuring
relations
simply represent
relations
in
the
subject
matter
(e.g.
of
succession,
cause
or
conditionality.)
Strictly
speaking,
the
relations
of
a
text
do
not
hold between
the
various
word sequences
of
which
the
text
consists.
Rather,
the
word
sequences
are
realizations
of
more
abstract
entities:
meanings
and
intentions
that
are
represented
by
those
word
sequences.
In
this
sense
all
of RST
is
pre-realizational.
since
it
makes
statements
about
how
such
meanings
and
intentions
are
structured
and
combined,
but
not
about
how
they
are
realized.
It
is
inconvenient
to
acknowledge
the
abstraction
on
every
mention,
so
we
will
generally say
that
relations
hold
between
spans
of
text.
but
the
distinction
between
the
abstract
entity
and
its
realization
always
remains.
Although RST identifies
the
nature
of
text
structuring
relations
as
functional,
it
does
not
presume any
particular
function.
Discovery
of
the
relations
and
their
functions
is
however an
empirical
matter.
In
research.
a
great
deal
of
misunderstanding
and
misrepresentation
of
language
has
come
from
assuming
that
the
sole
or
principal
function
of
language
is
informing,
and
that
it therefore
operates
as
a
message
passing
medium,
a
code.
This
agrees
with
the outlook
and
metaphors
of
our
culture
(content,
convey,
message,
language
as
a
conduit
[Reddy
79]),
but
it
does
not
stand
up
to
careful
examination;
see
for
example
the
abundant
exceptions
to
that
view
in
[Laridn
&
O'Malley
73].
The relations
in
fact
perform
a
diversity
of
functions;
some
are involved
with
informing,
but
many
perform
presentational
and
social
functions
with
little
informative
value.
If
one
sees
the function
of
text
as
predominantly
informing or
message-passing,
then
one
is led
to
a
view
that
text
structure
performs
predominantly
a
representational
f,
nction.
This
can
be
mitigated
by
a
broad definition
of
"message,"
but
not
corrected.

9
Another
alternative
is
to
assume
that
the
knowledge
of
text
structuring
relations
is
a
variety
of
lexical
knowledge,
e.g..
of
conjunctions.
This
assumption
would
be
most reasonable
if
the relations
were
always signalled
explicitly. (Something
close
to
this
assumption
can
be
arrived at
hy
beginning
one's
investigation
with
the
conjunctions
and
discovering
relations
from
them.)
In
RST
the
relations
are
not
identified with
any
particular
ways
in
which
they
are
represented.
Most
of
them,
but
not
all,
can
be
made
accessible
to
the
reader
by
conjunctions,
often
a
variety
of
conjunctions,
and
all
of
them
can
be
conveyed
in
other
ways,
including
being fully
implicit.
The
assumption
that
text
structuring
relations
are
lexical
is
an
alternative.
not
compatible with
RST.
9.
The
Number
of
Relations
--
The
set
of
text
structuring
relations
is
in
principle
open,
so
that
additional
previously
unused
relations
can
arise.
However,
the
frequency
of
creation
of
new
relations
is
extremely
low,
and
for
all
but
a
kernel
set
the
frequency
of
use
of
rare
or unknown relations
is
also
extremely
low,
so
that
text
in
a
culture
can
be
analyzed
virtually
entirely
in
terms
of
a
small
set
of
highly
recurrent
relations, the
knowledge
of
which
is
shared
in
the
culture.
The
relations
can
be
arranged
in
a
taxonomy,
with
the
particular
number
of
relations
reflecting
definitional
splitting
and
joining
of
taxonomic
categories. An
alternative
assumption
might
be
that
the
set
of
relations
used
in
a
particular
culture
and
situation
is
drawn
from
a
fixed
universal set,
possibly
not yet
fully
documented,
but
in
principle
not
expandable.
5
In
this
paper,
we
use
these
assumptions
to
help
characterize
RST and
to
clarify
the
analysis
at
points
where
the
assumptions
become
crucial.
The
assumptions
are
in
fact
built
into
the
mechanisms
of
RST.
Many
correspondences
of
these
assumptions
to
RST's
mechanisms
will be
evident
as
the
latter
are
described
below.
Space
limits
prevent
explaining
the
correspondence
here.
For
a
more
detailed
treatmeit
of
RST's
mechanisms,
see
[Mann
&
Thompson
88]
and
[Mann
&
Thompson
89].
The
use
of
RST
to
investigate
a
number
of
linguistic
issues serves
to
validate
RST's
assumptions.
Some
of
the
relevant
studies
are
described
below.
5A
paper
currently
in
preparation
refutes
this
alternative
assumption
by
showing
evidence
for
a
particular
mechanism
for
expanding the
set of
relations
in
culture-specific
ways.
!Mann
90

10
First,
RST
provides
a
general
way
to
describe
the
relations
among
organizational
elements
in
a
text,
whether
or
not
those
relations
are
grammatically
or
lexically
signalled.
Thus,
RST
is
a
useful
framework
for
relating
the
meanings of
conjunctions,
the
grammar
of
clause
combining,
and
non-signalled
parataxis
(For
discussion,
see
[Matthiessen
&
Thompson
89],
[Thompson
&
Mann
86]
and
[Thompson
&
Mann
87].)
Second,
descriptive RST
has
been
used
as
an
analytical
tool for
a
wide
range
of
text
types.
[Noel
86]
shows
how
it
can
be
used
to
characterize
news
broadcasts.
[Fox
871
demonstrates
how
explanations
of
the
choice
between
pronoun
and
full
NP
in
expository English
texts
can
be
derived
from
the organizational
structure
revealed
by
RST.
Third,
descriptive RST
lays
a
foundation
for
studies
in
contrastive
rhetoric.
Cui's
analysis
of
Mandarin
and
English
essays
[Cui
85]
is
an
example.
Fourth,
RST
has
proven to
be
useful
in
analyzing
narrative
discourse
as
well.
[Kumpf
86]
is
a
study
of
the
interlanguage
of
Japanese
and
Spanish speakers.
The
author
shows
that
RST
is
valuable
in
describing
the grammatical
and
rhetorical
properties
of
the
narratives
produced
by
these
speakers.
Finally,
RST
provides
a
framework
for
investigating
Relational
Propositions,
which
are
unstated
but
inferred propositions
that
arise
from
the
text
structure
in
the
process
of
interpreting
texts
(see
Section
7.1
and
[Mann
&
Thompson
86]).
Since
the
coherence
of
a
text
depends
in
part
on
these
Relational
Propositions,
RST
has
been
useful
in
the
study
of
text
coherence.

11
3.2
Terminology
and
mechanisms
of
RST
The
key
elements of RST
are
relations
and
spans.
6
Essentially,
the
relation
definitions
identify
particular
relationships
that
can
hold between
two
text spans.
A
text
span
is
any
portion
of
text
that
has
an
RST
structure
(and
thus
has
a
functional
integrity,
from
a
text-organizational
point
of
view),
or
that
is
realized
by
a
unit.
Units,
defined
in
Section
3.3.
are
typically
clauses.
(For
convenience
we
also
speak informally
of
the
region
of
text
that
realizes
a
text
span
as
being
a
text span.)
In
general,
a
text span
will
not
be
interrupted
by
another
text
span,
but
defining
text
spans
in
functional
terms
rather
than strictly
orthographic
terms
allows
for
interrupted
text
spans.
Section
8
discusses an
instance
of
an
interrupted
text
span
in
the
analysis
of
the
ZPG
letter.
The
notion
of
the
structure
of
a
text
is
defined
in
terms
of
the
network
of
relations
among
successively
larger
text
spans.
Relations
are
defined
to
hold
between two
non-overlapping
text
spans
called
the
nucleus
and the
satellite.
A
relation definition
consists
of
two
fields:
1.
Constraints:
including
a
set
of
constraints
on
the
nucleus,
a
set
of
constraints
on
the
satellite,
and
a
set
of
constraints
on
the
combination
of
nucleus
and
satellite.
2.
Effect:
including
a
statement
of
the
effect
that
plausibly
the
writer
was
attempting
to produce
in
employing
the
relation,
and
(derived
from
that
statement)
the
locus
of
effect,
identified
as
either
the
nucleus
alone
or
the
nucleus-satellite
combination.
6
For
a
more
formal
discussion
of
the
mechanisms
of
RST,
including
a
set
of
detailed
relation
definitions,
see
1Mann
&
Thompson
881.

12
We
can
see
how
these
fields
function
to
specify
a
relation
definition
by
taking
as
an
example
the definition
of
the
Evidence
relation,
shown
in
Figure
1.
1.
Constraints:
a.
Constraints
on
the
Nucleus:
The
reader
might
not
believe
the
nucleus
to
a
degree
satisfactory
to
the
writer.
b.
Constraints
on
the
Satellite:
The
reader
believes
the
satellite
or
will
find
it
credible.
c.
Constraints
on
the combination
of Nucleus
and
Satellite:
The
reader's
comprehending
the satellite
increases
his
or
her
belief of
the
nucleus.
2.
Effect:
a.
The
reader's
belief
of
the
nucleus
is
increased.
b.
Locus
of
the
Effect:
Nucleus.
Figure
1:
Definition
of
the
Evidence
Relation
The
Evidence
relation
is
appropriate
to
relate
two
text spans
one
of
which
(the
evidence
satellite)
provides
evidence
for
the
claim
put
forth
in
the
other
(the
nucleus).
As
an
example
we
can
consider
this
extract
from
a
letter
to
the
editor
of
BYTE
magazine: the
writer
is
praising
a
federal income
tax program published
in
a
previous
issue.
Here
Unit
2
provides
evidence
for
the
claim
in
Unit
1,
as
diagrammed
in
Figure
2.
1.
The
program
as
published
for
calendar
year
1980
really works.
2.
In
only
a
few
minutes,
I
entered
all
the
figures
from
my
1980
tax
returns
and
got
a
result
which
agreed
with
my
hand
calculations to
the
penny.
The
Effect
field
in
a
relation
definition
specifies
the
intended
effect
on
the
reader
of
that
particular
relation.
In
the
case
of
the
Evidence
relation,
for
example,
part
a.
of

13
1-2
evidence
1 2
Figure
2:
RST
Diagram
of
the
Tax
Program Letter
the
Effect
field
stipulates
that
the
writer
intends
that
the
effect
of
using
the
Evidence
relation
is
to
increase
the
reader's
belief
in
the
nucleus.
Thus
in
the
case
of
the
BYTE
magazine
example,
the
analysis
reflects
the
judgment
that
the
writer's
purpose
in
including
the satellite Unit
2
was
to
increase
the
reader's
belief in
the
claim
that
the
program
really
works.
Since
every
definition
has
an
Effect
field,
the
analyst
can
thus
provide
a
plausible
reason
the
writer
might
have
had
for
including
each
part
of
the
whole
text.
The
relation
definition
does
not
constrain
the
order
of
spans,
and
for
virtually
every
relation, both
of
the
possible
orders nucleus-satellite and
satellite-nucleus
are
found.
Closely
related
to
the
functionality
of
the
Effect
field
is
the
functionality
of
nuclearity.
Informally,
we
speculate
that
nuclearity
influences
the
way
the
reader
assigns
different
roles
to
different
parts
of
the
text.
If
the
satellite
gains
its
significance
through
the
nucleus,
the
writer
can
indicate,
by
the
inherent
nuclearity
of
the
relation
used,
that
the
nucleus
is
more deserving
of
response,
including
attention,
deliberation,
and reaction.

14
The
Locus
of
Effect
field
thus
allows us
to distinguish
between those
relations
whose
locus
of effect
is
the
nucleus from
those
whose
locus
of
effect
is
both the
nucleus
and the
satellite.
When
the
locus
of
effect
is
the
nucleus,
as
in
the
Evidence
relation,
nuclearity
represents
the
qualitative
differences
between
the
essential
and
the
inessential.
The
satellite
supports
the
nucleus,
but
does
not
contribute
to
it.
When
the
locus
of effect
is
both
nucleus
and
satellite,
a different
sort
of
function
is
performed:
the
relation
is
expressing
particular
characteristics
of
the subject
matter.
For
example,
a
relation
whose
locus
of
effect
is
both
nucleus
and
satellite
is
the
Circumstance
relation.
The
definition
of
the
Circumstance
relation
specifies
that
the
satellite
sets
a
subject
matter
framework
within
which
the
nucleus
can
be
interpreted.
The
intended
effect
is
simply
that
the
reader should
recognize
that
the
situation
presented
in
the
satellite
provides
the
framework
for
interpreting
the
nucleus;
its
locus
is
thus
both the
nucleus
and the
satellite.
As
an example
of
the
Circumstance
relation,
consider
this
extract
from
an
ad
for
computer
disks,
as
diagrammed
in
Figure
3.
1.
Cleaning
agents
on
the
burnished
surface
of
the
Ectype
coating actually
remove
build-up
from
the head,
2.
while
lubricating
it
at
the
same
time.
The
locus
of
the
effect of
the
Circumstance
relation
is
both
nucleus and
satellite:
the
intended
effect
is
that
the
reader
recognize
that
the satellite
(while
lubricating
it
at
the
same
time)
provides
a
framework within
which
to
interpret
the
nucleus
(Cleaning
agents
...
actually
remove
build-tp).

15
1-2
_circumstance
1 2
Figure
3:
RST
Diagram
Showing
the
Circumstance Relation
The
nuclearity
of
the
Circumstance
relation,
where
the
locus
of
effect
is
both
the
nucleus
and
the
satellite,
is
thus
quite
different
from
the
nuclearity
of
the
Evidence
relation,
where
the
locus
of effect
is
just
the
nucleus. Where
the
locus
of
effect
is
just
the
nucleus,
nuclearity
reflects
the
supporting
role
of
the
satellite;
where
the
locus
of
effect
is
both
the
nucleus
and
the
satellite, nuclearity
reflects
the
symbiotic
role
of
the
nucleus
and the
satellite
in
the
reader's
recognition
of
subject-matter
relationships.
Each
field
of
a
relation
definition
specifies
particular judgments
that
the
text
analyst
must
make
in
building
the
RST
structure.
Since
the
analyst
has
access
to
the
text,
has
knowledge
of
the
context
in
which
it
was
written,
and shares
the
cultural
conventions
of
the
writer
and
the
expected
readers,
but
has
no
direct
access
to
either
the
writer
or
other
readers,
judgments about
the
writer
or
readers
must
be
plausibility
judgments
rather
than
judgments
of
certainty.
That
is,
every
judgment
of
the
completed
analysis
is
of
the
form,
It
is
plausible
to
the
analyst
that
....
In
the
case
of
the
Effect
field,
for
example,
the analyst
is
judging
whether
it
is
plausible
that
the
writer
desires
the
specified effect
on
the
reader.

16
3.3
Steps
in
performing
an
RST
analysis
The
first
step
in
analyzing
a
text
is
to
divide
it
into
units.
Unit
size
is
arbitrary
in
RST:
in
principle
the
units
can
be
of
any
size
from
typical
lexical
items
to
entire
paragraphs
or
larger.
In
our
work
on
RST,
however,
we
have
found
it
useful
to
use
units
that
have
some
relatively
theory-neutral
functional
integrity.
We
have derived
interesting
results
from
considering
the
units
to
be
roughly
clauses,
except
that
clausal
subjects
and objects
and
restrictive
relative
clauses are
considered
parts
of
their
host
clauses
rather
than
separate
units.
78
The
units
we
are
working
with
here,
then.
are
typically
located
at
the
boundary
region
shared
by
relational
structure
and
syntactic
structure.
Larger
units
may
be
useful
for
various
other
purposes,
such
as
describing
the
overall
structure
of
larger
texts.
The
next step
is
to
identify
spans
and
relations,
working
either
from
the
top
down
(progressive
refinement)
or
from
the
bottom
up
(aggregation),
or
both.
as
deemed
convenient.
Again,
in
determining
what
relation
should
be
said
to
hold
between
two
7
The
rationale
for
this
is
that
these
clauses
are embedded
constituent
parts
in
other
structures
but
they
do
not
combine
with
other
clauses
in
terms
of
relational
structure.
Thus,
subject
and
object
clauses
enter
into
and
are
determined
by
the
transitivity
structure
of
the
clause
they
are
embedded
in.
In
contrast.
non-embedded
clauses
may
be
related
to
one
another
as
wholes
and
form
rhetorically
motivated
clause
combinations.
8
0n
this
basis
we
have
broken
down some
of
the
segments
of
the
ZPG
letter
into
parts
designated
with
subscripts
A,
B,
and
C
as
follows.
Subscript
A
always
designates the
first
part
of
the
segment.
6B:
answering
questions
6C:
and
talking
with
reporters
....
country.
7B:
we
had
no
idea
....
response.
I1A:
ZPG's
Urban
Stress
Test,
...
is
the
nation's
first
survey
...
cities.
1IB:
created
after
months
of
persistent
and
exhaustive
research.
14B:
we
urgently
need
your
help.
15B:
and
our
modest
resources
are
being
stretched
to the
limit.
19B:
we
can
act
to
take
positive
action
at
the
local
level.
21B:
both
elected
officials
....
study.
30B:
by
completing
the
enclosed
reply
form.

17
given
text
spans,
the
analyst
is
asking
at
each
point whether the
relation
definition
plausibly
applies.
Notationally,
we
represent
nuclei
under
vertical
lines
and satellites
at
the
ends
of
arcs,
as
shown
in
Figure
4,
which
shows
the
schema
representing the
top
level
organization
of
the
body
of
the
ZPG
letter.
While
the
schema
is
a
technical
device
in
the
more
elaborate
definition
of
RST,
here
we
can
think
of
it
simply
as
an
iconic
convention,
showing
how
units
and
relations
are
grouped.
In
the
case
of
nucleus-satellite
relations,
the
grouping
consists of
one
nucleus
and
all
of
its
satellites.
4-23
motivation
motivation
4-21
22
23
Figure
4:
Highest
Level
RST
Schema
Diagram
for
the
ZPG
Letter
As
in
other
kinds
of
analysis of
linguistic
structure,
RST
sometimes
yields
multiple
analyses for
a
text.
There
are
various
sources
of
this
miultipli(itv.
)Ile
of
which
is
ambiguity
of
discourse
structure.
We
present
a
single
hi ky-%is ,l,
.x:
-
are
comments
on
alternate
analyses of
the
text
in
Section
5.
....
...

18
The
set
of
relations
in
RST
is
not
a
closed
set.
Among
those
which
we
have
found
useful
are
the
following:
Nucleus-Satellite
Relations
Evidence
Justify
Antithesis
Concession
Circumstance
Solutionhood
Elaboration
Background
Enablement
Motivation
Volitional
Cause
Non-volitional Cause
Volitional
Result
Non-volitional
Result
Purpose
Condition
Otherwise
Interpretation
Evaluation Restatement
Summary
Multi-nuclear
Relations
Sequence
Contrast
Joint
Precise
definitions
for
all
these
relations,
and
of
the
conditions
under
which
they
can
be
applied,
can
be
found
in
[Mann
&
Thompson
89].
Here
we
will be
concerned
only
with
those
relations
that
figure
in
our
analysis of
the
ZPG
letter,
whose
definitions
can
be
found
in
the
Appendix.
The
role
of
subjective
judgment
in
the
theory
should
be
made
clear.
To account
for
communication
as
one
of
the
principal
functions
of
language,
a
linguistic
theory
must
be
functional,
in
the
sense
that
it
must
provide
representations
and
draw
conclusions
about
what
the
functions
of
particular
uses
of
language
are.
If
a
linguistic
theory
of
text
structure
is
to
be
functional,
judgments
about
the
functions
of
texts
and
text
parts
must
be
made
in
the
process
of
creating
and
testing
the
theory.
In
practice,
such
judgments
are
necessarily
subjective,
since
they
are
made only
by
human
beings
who
communicate,
on
the
basis
of
what
they
know
about their culture, their
society,
and
their
language.
This
kind
of
judgment
has
of
course
been
applied
frequently
in
developing
RST
(and
every
other
account
of language
function).
In
addition.
RST
employs
subjective

19
judgments
in
another,
more
controversial
role.
They
are
used
not
only
in
evaluating
an
analysis,
but
in
producing
it.
We use
this
approach
in
RST
because
it
is
an
effective
way
to
develop
functional
descriptions
of
text,
as
a
step
toward
ultimately
coming
to
understand
communication. Texts
are
complex
objects,
with correspondingly
complex
functional
descriptions.
To arrive
at
such
descriptions,
as
our
general
goals
require,
it
is
necessary
to
develop
and
combine
many
smaller
functional
descriptions.
9
The
specific
representation
of
this
kind
of
judgment
in
RST
is
that,
as
mentioned
above,
the
analyst
affirms
certain
statements
about
the
text
and
the
writer
as
plausible
rather
than
factual.
Related
to
this
approach
is
the fact
that
RST
relations
are
defined
without
referring
to
particular
textual
patterns;
the
Purpose
relation
is
defined
without
reference
to
in
order
to.
This
intentional
lack
of
explicit
hooks
makes
such
judgments
essential.
4
RST
Analysis
of
the
ZPG
Letter
The
original
ZPG
letter,
used
with
permission,
is
shown in
Figure
5.
We
have
segmented
it
for
analysis
as
shown
in
Figure
6.
The
integer
segment
numbers
are
based
on visual
subdivision, and
the
finer
segmentation
used
in
our
analysis
is
explained
below.
We
expect
that
the
holistic
structure
of
letters
is
specified
so
that
one of
the
alternatives
is
the
sequence
shown
in
Figure
7,
possibly
with
more
or
less
substructure
9
0f
course,
there
are
methodological
risks
in
approaching
the
problem
in
this
way,
risks
of
circularity,
divergence
of
analysis
from
actual
function,
nonrestrictiveness
of
the
theory,
vagueness,
indefiniteness
of
analytic
outcome,
etc.,
but
in
today's
linguistics,
descriptions must
be
produced
through
human
judgments
of
function
if
they
are going
to
be
produced
at
all.
These
issues
have
been
faced
and
some
progress made,
but
there
are
no
guarantees.

20
OBCERS
PAULR
EIIRLICH
November
22,
1985
KARIL
FROI4BOESE
EDXRD
BRANN
t
UDITH
JACOBSEN
JOYCE
LAMAN
ANDREWWIESSNER
Dear
Friend
of
ZPG:
EILEEN
PRATT
S-wy='
At
7:00
a.m.
on
October
25,
our
phones
started
.to
ring.
EDWIN
F.
LEACH,
11
"Calls
jammed
our
switchboard
all
day.
Staffers
stayed
late-
SUSAN
EBER
into
the
night,
answering questions
and talking
with
reporters
Cwt
iV
rCftfrom
newspapers,
radio
stations,
wire
services
and
TV
stations
SPONSORS
in
every
part
of
the
country.
lsaw
Asimw
Jessie
Berrnad
Gems
&X=tut,,
When
we
released
the
results
of
ZPG's
1985
Urban
Stress
Norman
E
Simau
Test
we
had
no
idea
we'd
get
such
an overwhelming
response.
Jim
SousMo
DavW
R.
Bfo%,
Media
and public
reaction
has been
nothing
short
of
incredible
.
Lester
R.
Bron
ROggf
C3sS
Ile
manE
Daly
At
first,
the
deluge
of
calls
came
mostly
from
reporters
KingseyDavis
eager
to
tell
the
public
about
Urban
Stress
Test
results
and
Uavne II.
Davis
Cat
ht.
t3,
from
outraged public
officials
who
were
furious
that
we
had
A.ne
I.Ehrlich
"blown
the
whistle"
on
conditions
in
their
cities.
Paul R
Ehrlkh
Otis
L.
Graham.J,.
Garrett
I ld.
Now
we
are
hearing
from
concerned
citizens
in
all
parts
of
Johnl
Holrn
Sam
G. Landfather
the
country
who
want
to
know
what
they
can
do
to
hold
local
u-r
Lovins
officials
accountable
for
tackling
population-related
problems
Shirley
Macuine
Ate
McCa0-
that
threaten
public
health
and
well-being.
Ian
L.
Mcllalt
Hclen
V.
Milliken
S"-n
MtX,
ZPG's
1985
Urban
Stress
Test,
created
after
months
of
Dicnw,
persistent
and
exhaustive
research
is
the
nation's
first
811h
ftclxmd
Lu~s
auing
survey
of
how
population-linked
pressures
affect
U.S.
cities.
=IIWPt'a.
It
ranks
184
urban
areas
on
11
different
criteria
ranging
from
IfSell
W.
Petero
Chfa1hiE
*,
crowding
and
birth
rates
to
air
quality
and
toxic
wastes.
Cluckes
E.
Scf4)M
Ridne-
Sha.
8
F
Sinne
The
Urban
Stress
Test
translates
complex,
technical
data
Siews
L. Wallcopedt
Ke
,
F
into
an
easy-to-use
action
tool
for
concerned
citizens,
elected
Ied"-N.
%XUMdWnd
officials
and
opinion
leaders.
But
to
use
it
well,
we
urgently
need
your
help.
Our
small
staff
is
being
swamped
with
requests
for
more
information
and
our
modest
resources
are
being
stretched
to
the limit.
Your
support
now
is
critical.
ZPG's
1985
Urban
Stress
Test
may
be
our
best
opportunity ever
to
get the
population
message
heard.
1601
CONNECTICUT
AVENUE.
NW
a
WSI
IINGTON.
DC
20009
a
(202)
332.2200

21
With
your
contribution,
ZPG
can
arm
our
growing
network
of
local
activists
with
the
materials
they
need
to
warn
community
leaders
about
emerging
population-linked
stresses
before
they
reach
the
crisis stage.
Even
though
our
national
government
continues
to
ignore
the
consequences
of
uncontrolled
population
growth,
we
can
act
to
take
positive
action
at
the
local level.
Every
day
decisions
are
being
made
by
local
officials
in
our
communities
that
could
drastically
affect
the
quality
of
our
lives.
To
make
sound
choices
in
planning
for
people,
ooth
elected
officials
and
the
American
public
need
the
population-
stress
data
revealed
by
our
study.
Please
make
a
special
contribution
to
Zero
Population
Growth
today.
Whatever
you
give
--
$25,
$50,
$100
or
as
much
as
you
can
--
will
be
used
immediately
to
put
the
Urban
Stress
Test
in
the
hands
of those
who
need
it
most.
Sincerely,
//Susan'Wber
Executive
Director
P.S.
The
results
of
ZPG's
1985
Urban
Stress
Test
were
reported
as
a top
news
story
by
hundreds
of
newspapers
and
TV
and
radio
stations
from
coast
to
coast.
I
hope
you'll
help
us
monitor
this
remarkable
media
coverage
by
completing the
enclosed reply
form.
Figure
5:
The
Original
Form of
the
ZPG
letter

22
Segment
I
ZERO
POPULATION
GROWTH
Segment
2
November
22,
1985
Segment
3
Dear
Friend
of
ZPG:
Segment
4
At
7:00
a.m.
on
October
25,
our
phones
started
to
ring.
Segment
5
Calls
jammed
our
switchboard
all
day.
Segment
6
Staffers
stayed
late
into
the
night,
answering
questions and
talking
with
reporters
from newspapers,
radio
stations,
wire
services
and
TV
stations
in
every
part
of
the
country.
Segment
7
When
we
released
the
results
of
ZPG's
1985
Urban
Stress
Test,
we
had
no
idea we'd
get such an
overwhelming
response.
Segment
8
Media
and
public
reaction
has
been
nothing
short
of
incredible!
Segment
9
At
first, the
deluge
of
calls
came
mostly
from
reporters
eager
to
tell
the
public
about
Urban
Stress
Test
results
and
from
outraged
public
officials who
were
furious
that
we
had
"blown
the
whistle"
on
conditions
in
their
cities.
Segment
10
Now
we
are
hearing
from
concerned citizens
in
all
parts
of
the
country
who
want to
know
what
they
can
do
to
hold
local
officials
accountable
for
tackling
population-related
problems
that
threaten
public
health
and
well-being.
Segment
11
ZPG's
1985
Urban
Stress
Test,
created
after
months
of
persistent
and
exhaustive
research,
is
the
nation's
first
survey
of
how,
population-linked
pressures
affect
U.S.
cities.
Segment
12
It
ranks
184
urban
areas
on
11
different
criteria
ranging
from
crowding
and
birth
rates
to
air
quality
and
toxic
wastes.
Segment
13
The
Urban
Stress Test
translates
complex,
technical
data
into
an
easy-to-use
action
tool
for
concerned
citizens,
elected
officials
and
opinion
leaders.
Segment
14
But
to
use
it
well,
we
urgently
need
your
help.
Segment
15
Our
small
staff
is
being
swamped
with
requests
for
more
information
and
our
modest
resources
are
being
stretched
to
the
limit.

23
Segment
16
Your
support
now
is
critical.
Segment
17
ZPG's
1985
Urban
Stress
Test
may
be
our
best
opportunity
ever
to
get
the population
message
heard.
Segment
18
With
your
contribution,
ZPG
can
arm
our
growing
network
of
local
activists
with
the materials they
need
to
warn community
leaders
about
emerging
population-linked
stresses
before
they
reach
crisis
stage.
Segment
19
Even
though
our
national
government
continues
to
ignore
the
consequences
of
uncontrolled population
growth,
we
can
act
to
take
positive
action
at
the
local
level.
Segment
20
Every
day
decisions are
being
made
by
local
officials
in our
communities
that
could
drastically
affect
the quality
of
our
lives.
Segment
21
To
make
sound
choices
in
planning
for people,
both
elected
officials
and the
American
public
need
the population-stress
data
revealed
by
our
study.
Segment
22
Please make
a
special
contribution
to
Zero
Population
Growth today.
Segment
23
Whatever
you
give
--
$25, $50,
$100
or
as
much
as
you can
--
will
be
used
immediately
to
put
the
Urban
Stress
Test
in
the
hands
of
those
who
need
it
most.
Segment
24
Sincerely,
Segment
25(handwritten
signature)
Segment
26
Susan
Weber
Segment
27
Executive
Director
Segment
28
P.S.
Segment
29
The
results
of
ZPG's
1985
Urban
Str-ess
Test
were
reported
as a
top
news
story
by
hundreds
of
newspapers and
TV
and
radio
stations
from
coast
to
coast.
Segment
30
I
hope
you'll
help
us
monitor
this remarkable
media
coverage
by
completing
the
enclosed
reply
form.
Figure
6:
The
Segmented
Form
of
the
ZPG
letter

24
or
grouping
than
is
shown.
The
body
and
P.S.'elements
of
this
sequence
are
specified
as
potentially
having
relational
structure.
Simeutan
Letterhea
The
RST
analysis
of
the
ZPG
letter
is
given in
Figure
8.
In
this
section,
we will
discuss
the
analysis
and the
rationale
for
the
judgments
we
have
made.
1 0
All
numbers
refer
to
units.
The
topmost
level
of
our
analysis
recognizes
certain
conventional
properties
of
a
letter
which
we
consider
not
to
be
part
of
its
relational
structure,
since,
as
pointed
out
in
Section
2,
they
are
part
of
the
holistic
structure;
they
are
included
as
part
of
what
we
know
about
the
form
of
letters
of
this
kind
in
our
culture.
Thus
1
is
a
letterhead
logo,
2
is
the
date,
and
3
is
the
greeting.
The
body of
the
letter
is
4
-
23,
which
has
an
RST
1
0
Recall
that
all
of
these
judgments
are
plausibility
judgments. Thus,
though
the
analysis
is
presented
as
if
it
were
Otruthu.
each
analytical
statement
in
it
should
be
read
as
It
is
plausible
that
the
writer
intended
....

25
UC4
U Iz
a, '
i
ft U
* I L
-A
2~
1 .1
Figre8:RSTDigrm
o
te/
.oy
f
h.ZG
ete

26
analysis,
to
be
discussed
below.
Units
24 27
constitute the
closing,
and
28
-
30
comprise
the P.S.,
whose
internal
structure
can
also
be
described
by
an
RST
analysis.
Before
discussing
the
body
of
the
letter,
let
us
briefly
discuss
the
RST
analysis
of
the
P.S.,
28
-
30.
The
analysis
reflects
the
afterthought
nature
of
a
P.S.
by
attributing
it
to
the
holistic
structure
associated with
letters,
and
not
taking
it
as a
discontinuous
element of
the
body
of
the
letter.
Internally,
we
have
considered
it to
consist of
a
Background
satellite
(29)
to
the
nuclear
30,
which
itself
is
composed
of
a
nucleus
(30A)
and a
Means
satellite
(30B).
According
to
its
definition,
the
Background
relation
is
appropriate
for
29
-
30:
the satellite
29
increases
the
reader's
ability
to
comprehend
an
element
in
the
nuclear
30
(namely
this
remarkable
media
coverage).
We
will
have
no
more
to
say
here
about
the
opening
1 -
3,
the
closing
24
-
27,
or
the
P.S..
Further
reference
to
the
text
will
be
to
the
body
of
the
letter,
4
-
23.
Let
us
now
turn
to
the
RST
analysis
of
the
body of
the
letter.
The
diagram
in
Figure
8
reflects
the
intuition
that
the
nuclear
unit
of
the
entire
text
(that
reached
by
tracing
from
the
top of
the
RST
structure
(here
the
node
labeled
4
-
23)
through
vertical
lines)
is 22.
This
is
appropriate
because
the
purpose
of
the
letter
is
blatantly
to
seek
contributions, and
it
is
in
22
that
the
appeal
is
stated
in
its
most
directive
fashion:
Please
make
a
special
contribution
to
Zero
Population
Growth
today.
Supporting the
nuclear
appeal
in
22
are
two
Motivation
satellites,
which
together
constitute
the
remainder
of
the text.
Thus,
the
RST
analysis
reveals
the
functional
structure
of
this
letter
to
consist
of
a
nuclear
request
accompanied
by
several
paragraphs
motivating
the
reader to
comply
with
the
request.

27
Taking
the smaller
Motivation
portion first,
we
note
simply
that
23.
assuring
readers
that
their
contribution
will
be
put
to
good
use,
is
the
final
sentence
in
the
letter
--
a
last-ditch
effort
at
motivating
readers
to
give
money.
The
larger
Motivation
chunk
is
the
entire
body of
the
letter
preceding
the
request,
comprising
4-21.
According
to
the
RST
diagram,
this
text
span
consists
of
a
nuclear
claim
11-16,
which
tells
readers
how
badly
their
help
is
needed,
flanked
by
two
pieces
of
evidence
for
that
claim,
4-10
and
17-21.
Let
us
now
consider
the
internal
structure
of
each
of
these
pieces
of evidence.
The
first
piece
of
evidence,
consisting of
4-10,
is
thematically
unified
around
the
public
reaction to
the
release
of
the
results
of
the Urban
Stress
Test.
That
is,
this
span
provides
evidence
for
the
claim
that
readers'
help
is
needed
by
detailing
how
ready
the
public
is
for
the
ZPG message
that
the
Urban
Stress
Test
provides.
The
nucleus
of
this
segment
is
7.10, which
describes
the
unexpected public
response
to
the
Test
results.
As
Background
to this
nuclear
claim,
we
find
the
narrative
sequence in
4-8,
describing
the events
of
the
day
the
Test
results
were
released. These
narrative
Units
4-6
are
related
to
each
other
by
the multi-nuclear
relation
Sequence.
Note
that
within
this
sequence,
6
can
be
subdivided
into
a
nuclear
6A
plus
two
Elaboration
satellites,
6B
and
6C.
Looking
more
closely
now
at
the
nuclear
span
of
this
first
Evidence
satellite,
7-10,
we see
that
this
span
consists of
the
nuclear
pair
7-8,
and
an
Elaboration
satellite.
Units
7-8
are
related
to
each
other
by
the
bi-nuclear relation
Restatement.
That
is,
8
restates
the
content
of
7,
the
idea
that
the
public
response
has
been

28
overwhelming
and
incredible.
(Note
that
7
itself
consists
of
a
nuclear
7B
preceded
by
a
when-clause
in a
Circumstance
relation.)
Finally,
9-10,
the elaboration
of
7-8,
generalizing the
reactions
during the
first
month
after
the
release
of
the
Test results,
are
related
to
each
other
by
the
bi-nuclear
relation
Contrast.
That
is,
9
contrasts
with
10
in
that
9
gives
the
initial reactions,
coming
from
reporters
and
outraged
public
officials,
while
10
gives
the
less
immediate
and
more
moderate
reactions
from
concerned
citizens
in
all
parts
of
the
country.
The
first
piece
of
evidence,
then,
for
the
claim
that
readers'
help
is
needed,
is
the
portion
of
the
text
describing
the
positive
reactions
to
the results
of
the
Urban Stress
Test.
The
second
piece
of
evidence
for
this
assertion
that
help
is
needed
is
the
span
17-21,
thematically
discussing
the
role
that
the Test
can
play
in
raising
public
consciousness
about
population-related
problems
in
cities.
The
nucleus
of
this
span
is
17,
which claims
that
the
Urban
Stress
Test
may
be
the
best
way
to
disseminate
the
population
message;
it
is
followed
by
two
Elaboration
satellites.
The
first
of
these
Elaboration
satellites,
18-19,
consists
of
two
spans
in
a
Restatement
relation:
19
restates
(we
can
take
positive
action
at
the
local
level)
what
18
conveys
(warn
community
leaders
about
population-related
stresses).
Unit
19
itself
consists
of
a
Concesion
satellite
19A
and
a
nuclear
19B.
The
second of
the
Elaboration
satellites
is
20-21.
Unit
20
is
in
a

29
Solutionhood
relation with
21.
and
21A
is
in
a
Solutionhood
relation
with
21B.
Thus,
the
problem
in
20,
that
local
officials
are
faced
with
decisions
that
affect
us. is
solved
by
21,
providing
them
with the
data
from
the
Test
to
enable them
to make
sound
choices.
But
21
also
consists
of
a
problem
and its
solution:
21A
presents
the
problem
of
making sound
choices,
and
21B
presents
the
solution,
which
is
to
get
the
data
to
elected officials
and the
American
public.
We
have
now
described
both
of
the
Elaboration
satellites
in
the
Motivation
span
4-21.
Before
going
on
to
the
nucleus
of
this
Motivation
satellite,
let
us
pause
briefly
to
comment
on
a
feature
of
Unit
21,
namely
its
indirectness.
As
we
have
just
suggested,
the
RST
analysis
claims
that
it
is
plausible
to
analyze
21
as
presenting
a
solution
to
the
problem
in
20
of
local
officials
making
decisions
that
affect
our
daily
lives.
But
a
careful
reading
of
21
shows
that
it
does
not
literally present
such
a
solution,
since
all
it
says
is
that
elected
officials
and
the American
public
need
the
population-stress
data.
However
it
is
clear
in
understanding
21
that
the
writer
is
suggesting
that
when
the
officials
have
the
data
it
helps
in
solving
their
problem
of
making
sound
decisions.
Analysis
must
take
many
kinds
of
indirectness
of expression
into
account,
including
indirect
speech
acts,
various
kinds
of
hedging,
metonymy
and
many
other
ploys.
Now
let's
turn
to
the
nucleus
(11-16)
of
the
Motivation
satellite
whose
two
E'idence
satellites
we
have
just
considered. Units
11-16
are
related
by
the
Concession
relation:
11-13
form
a
Concession
satellite to
the
nuclear
14-16.
Since
this
example
of
Concession
is
not
as
straightforward
as
that
seen
in

30
19A-10B,
let's
pause
briefly to
justify
it.
As
discussed
in
[Thompson
&
Mann
861
and
[Mann
&
Thompson
89],
the
Concession
relation
can
be
fruitfully
thought
of
as
involving
apparent
incompatibility
but
actual compatibility. The
definition
of
Concession
(see
Appendix
I)
includes
the
provision
that
the
writer
intends
the actual
compatibility
of
the satellite
and
the
nuclear spans to
increase
the
reader's
positive
regard
for
the
nucleus.
In these
terms,
the span
11-16
can
be
seen
as
contrasting
11-13,
describing
the
positive
attributes
of
the
Urban
Stress
Test, with
14-16,
which
point
out
the
desperate
financial
straits
of
ZPG. These two
situations
are
potentially
incompatible,
since
a
lack
of funds
decreases
the
value
of
the
Urban
Stress
Test.
However,
the
writer
views
the
two
situations
as
compatible;
they
are
compatible
if
readers respond
with
the
needed
funds.
And
she
hopes
to
increase
the
reader's
positive
regard
for
the
nuclear
span
describing
the desperate
financial
straits
by
getting
the reader
to
recognize
the
compatibility
as
well.
We
hope
to
have
convinced
you
that
11-13
are
plausibly analyzed
as
constituting
a
Concession
satellite
to
14-16.
Within this
Concession
satellite
itself,
11-13,
13
is
the
nucleus,
with
11-12
as
an
Background
satellite.
That
is,
11-12
provide
a
basis
for
understanding
the
statement
about
"translation"
in
13.
Units
11-12,
in
turn,
are
in an
Elaboration
relation, with
12
providing
details
of
11,
and
liB,
the
participial
clause,
is
a
further
elaboration
for
11A.
The
nucleus
of
the
Concession
relation
in
11-16
is
14-16.
Within
this
span,
15
is
the
nucleus
and
14
and
16
are
satellites,
each
in
a
Non-volitional
Result
relation.

31
That
is.
14
we
need
your
help and
16
your support
is
critical
are
plausibly
intended
by
the writer
as
results
of
the
situation
described
in
15.
Units
14A-B
can
be
further
analyzed
in
terms
of
the relation
Purpose.
Finally,
15A
and
15B
together
form a
Joint
Schema;
that
is.
they
jointly
perform
the
double
role
described
for
15,
and
none of
the
other
RST
relations
holds
between
them,
as
specified
by
the
definition
of
Joint.
1 1
Our
description
of
the
RST
analysis
for
the
ZPG
text
is
now
complete.
The
analysis
shows
that
the
body
of
the
ZPG
letter
can
be
understood
as
a
request
for
donations
(22),
preceded
by
a
lengthy
portion
(11
paragraphs)
motivating
readers
to
comply
with
this
request.
Within
this motivation
section,
we
have
seen
that
a
nuclear
claim
that
help
is
urgently
needed
is
flanked
by
two
pieces
of
evidence
for
this
claim.
One
of
these
pieces
of
evidence
asserts
that
the Urban
Stress
Test
has
been
very
well
received,
and
the
other
claims
that
the Urban
Stress
Test
can
be
useful
for ZPG
action
at
the
local
level.
Now
we
can
review
the
ways
ln
which
this
particular
analysis
represents
the
general
assumptions identified
in
Section
3.1.
The
assumption
of
Organization
(#1)
is
obviously
represented
by
the
structural
analysis.
The
assumption
of
Unity
and
Coherence
(#2)
is
represented
by
the
fact
that
for
each
of the
relational
parts
(the
body
and
the
P.S.)
every
part
is
incorporated into
a
single
connected
analysis.
That
unity
and
coherence
arise
from
imputed
function
(#3)
'lRecall
that
in a
more
formal
presentation
of
RST there
is
another
mechanism.
called
the
schema,
which
is
useful
for
specifying
how
relations
can
be
combined
on
a single
nucleus,
and for
handling
various
exceptional
structures.
In
that
treatment
there
is
a
Joint
schema
but
no
Joint
relation.

32
is
represented
by
the
requirement
that
the
analyst
confirms
the
fit
of
the
relation
definitions
used
in
the analysis, including
the
Effect
field,
according
to
which
the
analyst
imputes
to
the
writer
an
intention
of
effect
for
each
relation
employed.
Roles
in
terms
of
intended
effects
are
thus
assigned
to
every
part
of
the
text.
(The
assumption
that
the
nature
of
text
structuring
relations
is
functional
(#8)
is
fulfilled
in
the
same
way.)
The assumptions
of
hierarchy
(#4)
and homogeneity
of
hierarchy
(#5)
are
represented
in
RST's
mechanisms,
which
produce
a
hierarchy
by
using
the
same
mechanism
at
every
level.
The
assumptions
on
relational
composition
(#6)
and
the
dominance
of
asymmetry
(#7)
are
fulfilled
in
the
analysis,
since
of
the
31
terminal
units,
23
stand
in
a
nucleus-satellite
relation,
6
in
a
multinuclear
relation
and
only
2
in a
Joint
structure.
In
addition,
there
are
10
nonterminal
nucleus-satellite
relations
and
no
multinuclear
ones.
Finally,
the
assumption
that
in
practice
the
number
of
relations
is
small
(#9)
is
seen
in
this
analysis,
which
uses
14
different relations
to
accomplish
30
links.
A
list
of
these
relations
is
in
Figure
9.
5
Alternative
Analyses
It often happens
that
a
text
has more
than
one
analysis;
it
is
a
normal
and
predictable
outcome,
given
the
way
that
RST
is
defined.
We
and
others
have
had
the
experience
of
giving
the
same
text
to
several
analysts,
who
then
created
differing
analyses,
sometimes
more
than
one
from an
individual analyst.
There
are
several
qualitatively
different
causes
of
this
multiplicity:

33
Background
Means
Motivation
Evidence
Concession
Elaboration
Non-volitional Result
Sequence
Circumstance
Contrast
Solutionhood
Joint
Restatement
Purpose
Figure
9:
The
Relations
of
the
ZPG
Letter
1.
Boundary
Judgments
--
results
of
forcing
borderline
cases
into
categories.
2.
Text
Structure
Ambiguity
--
comparable
to
many
other
varieties
of
linguistic
ambiguity.
3.
Simultaneous
Analyses
--
multiple compatible
analyses
(see
[Ford
871
for
some
discussion).
4.
Differences
Between
Analysts
--
especially,
differing
plausibility
judgments.
5.
Analytical
Error
--
especially by
inexperienced
analysts.
There
is
a
well-known
phenomenon
associated
with grammatical
ambiguity,
in
which people
initially
regard
a
construct
as
unambiguous,
and
only
later
recognize
that
there
are
other
analyses.
This
fixation
on
particular
analyses
arises
in
RST
as well.
This
is
seen
when
several
analysts
analyze
the
same
text
and
then
accept
each
other's
analyses.
With
experienced
analysts,
multiplicity
of analyses
represents
primarily
simultaneous
analyses
and
text
structure
ambiguities.
The
particular
role
of
the
analyst
causes
bizarre
analyses
to
be
legitimately
rejected, and
so
the
actual
levels
of
ambiguity
are
much
lower
than
experience
with
formal
grammatical
analyses
woiId
lead
one to
expect.
Multiplicity
of
RST
analyses
is
normal,
consistent
with linguistic
experience
as a
whole,
and
is
one
of
the
kinds
of
pattern
by
which
the
analyseb
are
informative.

34
By
way of
illustration,
we
take
one
such
case
from
the
ZPG
letter:
it
concerns
the
appropriate
analysis
of
the
relation
of
11-13
to
14-16.
In
Section
8
we
justified
labeling
this
relation
Concession,
acknowledging
that
this
analysis
might
need
justification.
Indeed,
this
relation
could
be
thought
to
be
an
instance
of
Background.
However,
we
argue
that
Concession
is
more
plausible.
Recall
that
11-13
characterize the
Urban
Stress
Test,
while
14-16
bemoan
the
lack
of
resources
and
issue
the
first
plea
for
financial
support.
Before
we
consider
the
multiple
analysis
for
these
two
spans,
we
briefly
note
that
there
are
two
analyses for
the
relation
between
11-13
and
14-16
that,
although
initially
attractive,
are
not
real
options.
These are
Contrast
and
Antithesis,
since
the
two
spans
do
not
serve
to
compare
two
situations
and
contrast
them
on
one
or
more
points
of
difference,
as
required
by
the definitions
of
these two
relations.
However,
it
is
certainly
plausible
that
11-13
be
analyzed
as
providing background
to
14-16,
according
to
our
definition of
Background,
since
it
could
be
seen
as
containing information
necessary
for
the
interpretation
of
14-16.
See
Figure
10
for
a
diagram
of
this
alternative.
Although
this
is
a
plausible
analysis,
we
do
not
take
backgrounding
to
be
the
primary
function
of
11-13
in
this
highly
manipulative
text.
Rather,
as
suggested above,
we
claim
that
11-13
are
serving to
highlight
the
appeal
for
funds
in
14-16
by
pointing
out
all
the
virtues
of
the
Urban
Stress
Test
as
potentially
incompatible
with
using
it
--
incompatible
unless
the
needed
funds
are
donated.
This
case
of
multiple
analyses
seems
to
be
a
genuine
case
of
what
we call
"Simultaneous
Analyses";
that
is,
there
are
two
compatible
analyses,
but
one
of
them

35
appears,
to
the three
of
us
at
least,
to
be
more
plausible
in
terms
of
the
perceived
overall
goals
of
the
writer.
12
1 -16
background
11-13
14-16
Figure
10:
An
Alternate
Analysis
of
Units
11
through
16
6
Consequences
and
Inferences
from
the
Analysis:
Rhetorical
Patterns
6.1
Centrality
at
the
Whole-Text
Level
One
of
the outcomes
of
an
RST
analysis
is
identification
of
a
portion
of
the
text
that
represents
the
essence
of
the
text
as
a
whole.
It
is
called
the
Comprehensive
Locus
of
Effect.
It
is
arrived
at
by
pruning
the
structure
diagram
as
follows:
Starting
at the
top
(whole-text)
node,
trace
down
to
each
nucleus.
For
each
relation
linking
nucleus to
satellite,
if
the
relation
is
defined
to
have
a
locus
of
effect
that
is
the
nucleus only,
remove
the
satellite.
Trace
down
repeatedly
in
the
same
way
to
all
terminal
units.
The
resulting
"text",
with
some
satellites
deleted,
is
the comprehensive
locus
of
effect.
In
our
experience
this
resulting
"text"
remains coherent
and
represents
a
kind
of
120iir
thanks
to
Peter
Fries
for
discussion
of
this
point.

36
ideational
essence
of
the larger
text.
It
is
formally
representative
in
the
sense
that
the
whole
text
is
considered
in
deriving
it.
For
the
ZPG
letter
body,
the
comprehensive
locus
of
effect
is
simply
segment
22:
Please
make
a
special
contribution
to
Zero
Population
Growth
today.,
which
certainly
reflects
the
judgment
that
this
letter
is
in
essence
an
appeal
for
funds.
For
the
P.S. the
comprehensive
locus
of effect
is
the
second
sentence:
I
hope
you
'1
help
us
monitor
this
remarkable
media
coverage
by
completing the
enclosed
reply
form.
,
which reflects
the separate
appeal
being
made
in
the P.S.
As
one
of
the
outcomes
of
a
whole-text
analysis,
we
can
have
some
confidence
that
the
comprehensive
locus
of
effect
does
not
misrepresent
the
text
as
an
informal
summary
might,
for
example
by focusing on
only
one
portion.
6.2 A
Presentational
Pattern
The
RST
analysis
brings
out
an
interesting
pattern
in
the
organization
of
the
ZPG
text.
As
Figure
8
shows,
there
are
three
places
in
the
analysis
where
a
nucleus
is
flanked
by
two
satellites
of
the
same
relation;
these are
reproduced
in
Figure
11.
This
pattern
is
readily
visible
only
with
an
analytical
tool
such
as
RST
offers,
in
which
there
is
a
distinction
between
nuclear and
satellite portions
of
the text.
This
pattern
is
somewhat
unusual
among
the
more
than
400
texts
we
have
analyzed.
Without
many
more
examples
of
this
Satellite-Nucleus-Satellite
pattern,
we
cannot
be
sure
whether
it
reflects
properties
of
the
appeal-letter
genre
or
habits
of
the
author
of
this
particular
letter,
but
in
either
case
it
seems
to
reveal
the
repetitive
mode
in
which

37
4-23
4-21
22
23
4-10
11-16
17-21
14-16
aoa-volitiona
cam
nmn-volitional cause
14
iS
i
Figure
11:
Satellite-Nucleus-Satellite
Patterns
in
the
ZPG
Text
this
appeal
letter
is
written.
That
is,
this
Satellite-Nucleus-Satellite
pattern
highlights
our
feeling
as
readers
that
the
letter
is
strongly
organized
around
a
mode
of
leading
up
to
the point,
stating
the
point,
and
driving
it
home.
7
Consequences
and
Inferences
from
the
Analysis:
Relations,
Grammar
and
Lexis
7.1
Relations,
Relational
Propositions
and
Explicit
Signals
Another
kind
of
consequences
of
the
text
structure,
labeled
"relational
propositions"
in
[Mann
&
Thompson
86],
involves
communication
of
information
in
the
relational
structure
itself.
The
relations
can
communicate
for
the
writer,
just
as
clause
structure
or
words
can.
Sometimes
this
communication
is
entirely
implicit;
sometimes
it
is
signalled
in
w~rious
ways.
(By
a
signal
of
a
relation
we
mean
any
single
lexical,
morphological
or
syntactic
construct
that
regularly
occurs
in
conjunction
with
the
relation,
and
that
can
be
taken

38
to
represent
(perhaps ambiguously)
the
presence
of
the
relation.
There
are
some
complex
patterns
that
are
outside
of
this
definition,
and
some
possibilities
for
inferential
recognition
of
relations
that
are
outside
of
both
of
the
above.)
For
example,
segment
20
talks
about
decisions
by
local
officials
that
could
drastically affect
the
quality
of
our
lives.
Segment
21
talks
about
these
officials
having
the
results
of
the
study.
In
the
analysis
there
is
a
Solutionhood
relation
between these two.
The
satellite
(20)
presents a
problem
of
decision-making
(or
our
living
with the
results),
for
which
having
the
data
is
a
(partial)
solution. Mediating
between
these
is
the
implicit
idea
that
the
quality
of
the
decisions
may
be
improved
if
they
are
informed
decisions.
A
plausible
inferential
path
would say
that
if
the
officials
have
the
data,
then
they
may
use
it
in
making
decisions,
and
that
if
they
use
it,
the
quality
of
the
decisions
may increase,
and
that
if
the
quality
of
the
decisions
increases
then
the
quality
of
cur
lives
may
increase.
Having
the
data
is
thus
a
(partial
and
uncertain)
solution to
the
problem
of
quality
presented
in
segment
20.
This
claim
of
solutionhood
is
the
relational
proposition;
it
is
conveyed
by'
the
solutionhood
relation
itself,
and
is
not
explicitly
signalled.
It
is
this solutionhood
that
makes
the
satellite
relevant;
even
though
the
claim
is
obviously
quite
tenuous,
if
it
were
denied
(say,
elsewhere in
the
text)
then
segment
20
would
become
a
nonsequitur
and
the
coherence
of
the
text
as
a
whole
would
thus
be
broken.
Relational propositions
have
a
status
comparable
to
that
of
the
explicit
assertions
of
the
text.
The
writer
is
responsible
for
their
veracity,
and
can
be
quoted
as
claiming
their
truth.

39
Relational propositions
affect
our
interpretation
of
texts
in
two
ways.
They
help
explain
the
significance
of
various
kinds
of
structure-indicating
signals
when
they
are
present.
In
the
absence
of
signals,
they indicate
the
basis
of
coherence
of
texts
and
specify
part
of
what
the
text
communicates.
For
some
texts,
recognizing
this
kind
of
implicit
communication
is
crucial
to
explaining the
text
as
a
means
of
communication.
A
corollary
to
this
is
that
RST
predicts
that
although
some
relations
may
be
signalled
by
conjunctions
or
connectives
of
various
sorts,
the
relations
will
be
recognized
even
if
they
are
unsignalled.
Strikingly,
most
of
the relations
in
the
ZPG
text
are
unsignalled.
This
is
strong
confirmation
of
the
assumptions
of
RST:
a
text
is
understood
to
cohere
in
the
ways
that
it
does
largely
by
virtue
of
its
relational
structure
rather than
by
virtue
of
overt
markers
signalling
relations
among
its
parts.
In
other
words,
connectives
are
better
thought
of
as
guiding
the
interpretation
of
a
text
than
as
necessary
signals
of
relations.
Thus
there
are
only
eight
relational
signals
in
the
text.
Let's
briefly
consider
each
of
them.
The
first
is
the
when
at
the
beginning
of
7A,
marking the relation
between
7A
and
7B
as
one
of
Circumstance.
Next
is
a
pair
of adverbs
that
together
signal
the
Contrast
relation
between
9
and
10.
As
we
pointed
out
in
the
analysis,
9
gives
the
initial
and
stronger
reactions,
while
10
gives
the
less
immediate
and
more
moderate reactions
to
the
Urban
Stress
Test
news.
As
signals,
we
find
At
first
at
the
beginning
of
9
and
Now
at
the
beginning
of
10.
The
next
two
explicit
signals
are
signals
of
concession.
The
first
is
But
at
the

40
beginning
of
14.
Recall
that
11-13
and
14-16
are
in
a
Concession
relation,
which
we
justified
at
some
length
in
Section
4
above.
The
second
signal of concession
is
the
Even
though
in
19A,
marking
19A
as
a
Concession
satellite
to
19B.
This
situation
nicely
illustrates
our
assumption
that
relations
are
perceived
to
hold
between
parts
of
a
text
largely
independently
of
the
signals
used.
A
relation
may
be
unmarked, or,
as
in
this
case,
it
may
be
marked
by one
of
a
variety
of
signals.
A
Concession
relation
is
typically
either
marked
on
the
satellite with
a
"concessive
conjunction"
such
as
although,
or
marked instead
on
the
nucleus
with
a but.
The
claim
made
in
[Thompson
&
Mann
86],
that
concession
is
a
matter
of
discourse
relations
rather
than
a
matter
of
a
certain
type of
"subordinate"
clause,
is
supported
by
the
ZPG
text.
After
the
But
in
14A,
there
is
an
infinitive
purpose
clause,
marked with
a
to.
The
next
relational
signal
in
the
text
is
the
To
in
To
make
sound
choices
in
21A,
which
signals
the
Solutionhood
relation
between
21A
and
21B. 21A
talks
about
making
sound
choices
and
21B
talks
about
getting the population-stress
data
to
those
who
make
these
choices.
Thus
the
satellite
21A
presents
a
problem
of
decision-making,
for
which
having
the
data
is
a
(partial)
solution. The
infinitive
To
make
sound
choices
sets
up
the
problem
to
which
the
following
main
clause
presents
the
solution.
This
example
provides
interesting
confirmation
of
our
claim
that
relations
and
signals
do
not correspond
in
any
simple
way. Note
that
in
terms
of
grammatical
form,
the
infinitive
clause
in
14A
(to
use
it
well)
and
that
in
21A
(to
make
.ound
choices
...)
are
identical.
Yet
the
first
one
accompanies
a
Purpose
relation,
and
the
second
one
accompanies
a
Solutionhood
relation.
That
is,
analyzing
21A-21B
as
in
a

41
Solutionhood
relation,
rather
than
as
in
a
Purpose
relation,
is
consonant
with
our
perception
of
the
writer's
goal
to
motivate the
reader to donate
money:
it
is
more
plausible
that
the
writer
is
bringing
up
the
idea
of
officials
making
decisions
that
affect
our
lives
as
a
problem,
which
dissemination
of
the test
data
can
help solve,
than
that
she
is
claiming
that
officials
and
the
public
need
the
data
for
the purpose
of
making
sound
choices.
Such
an
argument
cannot
readily
be
made
for
the relation
between
14A-B.
We
will
return
to
20A-B
shortly,
as
it
illustrates
another
consequence
of
RST.
Finally,
in
the
P.S.,
there
is
a
Means
satellite
signalled
by
by
in
30B.
Thus,
of
all
the
relations
in
our
analysis,
as
shown
in
Figure
8,
only
eight
of
them
are
explicitly
signalled.
Most
of
the
interpretation
of
the
coherence
of
this
text.
then,
is
achieved
by
inference.
The
abundance
of
unsignalled
relations
highlights
the
importance
of
the
differences
between
text
understanding,
which
involves
recognition
of
text
structure
and
relations,
and
a
symbol
decoding
task.
Recognizing
relations
requires
that
the
reader
make
judgments
about
the
writer,
including
judgments
of
the
plausibility
of
intentions.
Symbolic
decoding proceeds on
the
basis of
the
conventional
import
of symbols
and
their
compositions,
without
judging
such
factors.
This
suggests
that
much
ongoing
work
that
seeks
to
develop
a
formal
compositional
semantics
of
language
will
be
permanently
incomplete
with respect
to
coherent
texts.

42
7.2
"Subordination"
Another
consequence
of
our
analysis of
the
ZPG
text
in
terms
of
RST
is
the
insight
we
can
gain
into
the
issue
of
"subordination".
As
suggested
in
[Matthiessen
&
Thompson
89],
one
kind
of
what
has
been
studied
as
grammatical
"subordination"
is
perhaps
better
viewed
as
a
clause-level
and
partially
grammaticized
reflex
of
nucleus-satellite
patterns
found
pervasively
in
text.
Following
[Halliday
851,
Matthiessen
and
Thompson
propose
to
abandon
the
term
"subordination"
and
to
make
a
distinction
between
hypotaxis
and
embedding.
Hypotaxis
includes
the
category
of
"subordinate"
clauses
that
are
generally
thought
of
as
adverbial
clauses
in
English
grammar:
clauses
expressing
time,
reason, cause,
condition,
etc.
Embedding,
on
the
other
hand,
includes
essentially
restrictive
relative
clauses,
subject
and
object
clauses,
and
clausal
complements
to
verbs
and
adjectives.
This distinction
turns
out
to
have
far-reaching
consequences
as
we
consider
the
relationship
between
"subordination"
and
text
structure,
since
hypotaxis,
but
not
embedding,
seems
to
play
a
role
at
the
clause
level
very
similar
to
that
of
the
relations
IL
of
RST
in
text
organization
in
general.
In
other
words,
hypotactic
clauses
seem
to
be
clause-level
satellites
in
the
relational
structure
of
the
text,
no
different
in
function
from
other
satellites
encompassing
many
clauses.
To
fully
support
this
position
is
beyond
the
scope
of
this
paper
(for
discussion,
see
[Matthiessen
&
Thompson
89]);
however,
in
the
ZPG
text, there
are several
interesting
pieces
of
support
for
this
view.
First,
note
that
the
text
contains
two
Solutionhood
relations,
which
happen
to
be
adjacent
to
each
other,
the relation
between
20-21,
and the
relation
between

43
21A-21B.
We
have
commented
on
both
of
them:
the first
provided
an
illustration
of
relations
without
signals,
while
the
second
provided
an
illustration
of
one
grammatical
form
(an
infinitive)
signalling
both
a
Purpose
relation
(14A-B)
and a
Solutionhood
relation
(21A-B).
Now,
in
the context
of
"subordination",
we
can
tie
these
arguments
together
to
show
that
the
hypotactic
clause
To
make
sound
choices...
in
21A
is
functioning
in
the
relational
structure
of
the
ZPG
text
in
precisely
the
same
way
as
the
non-"subordinate"
sentence
in
20:
both
are
Solutionhood
sateliites
to their
nuclei.
Second,
the
text
contains a
number
of examples
of
embedding.
None
of
these
perform
text-organizing functions
in
the
same
way
as
do
the
hypotactic
clauses,
as
just
illustrated
for
the
infinitives
expressing
Purpose
and
Solutionhood
relations.
For
example,
to
continue
the
discussion
in
terms
of
grammatical
infinitives,
notice
that
9
contains
the
verb
phrase
eager
to
tell
the
public....
The
infinitive
in
this
verb
phrase,
however,
unlike
those
in
14A
and
21A,
is
not functioning
in
the text
to
signal
any
of
the
relations
of
RST
(see
Appendix);
rather
it
is
a
grammatical
complement
to
the adjective
eager,
and
as
such
is
part
of
the
predicate
to
which
eager
belongs,
which
itself
is
embedded
as
a
truncated
relative
clause
modifying
reporters.
The
same
can
be
said
for
each
instance
of
embedding
in
this
text.
To
take
just
one
further
example,
none
of
the
relative
clauses
in
the
text
play
a
role
in
the
relational
structure
of
the text.
For
instance,
9
contains
the
relative
clause
who
were
furious
that
we
had
"blown
the
whistle"
....
Such
a
relative
clause
tells
us
something
about
the
public
officials
calling
the
ZPG
office,
but
it
does
not
relate
portions
of
text
in
the
same
way
that
the RST
relations
do.

44
In
other
words, when
we
consider
the
function
that
these
embedded
clauses
have
--
such
as
forming
complements
to
adjectives
and verbs,
or
serving
to
modify
noun
phrases
-- we
find
that
these
types
of
functions
are
not
needed
in
the
description
of
relations
between
larger
units.
The
RST
analysis
of
the
ZPG
text,
then, supports
the
general
finding
that
of
the
two
types
of
"subordination"
which have
been
identified
in
the
literature,
only
one
of
them, hypotaxis,
can
be
understood
as
involving
the
relations
found
at
higher
levels
of
text
organization.
7.3
Concessives
Finally,
the
RST
analysis confirms
earlier
findings
relating
to
the
Concession
relation.
[Thompson
&
Mann
86]
point
out
that,
although
the literature
on
concession
concerns
itself
largely
with
the
meanings
of
certain conjunctions
such
as
although,
concession
can
be
insightfully
viewed
in
terms
of
discourse
relations
rather
than
simply
in
terms of
the
meanings
of
conjunctions.
As
noted
in
Sections
4
and
7.1,
the
two
examples
of
the
Concession
relation
in
the
ZPG
text
are
signalled
in
different
ways:
one
with
an
adverbial
clause
marked with
even
though
and
one
with
a
main clause
marked
with
but.
Viewing
concession
as
a
meaning
of
certain
types
of
"subordinate"
clause
conjunctions
would
miss
the functional
similarities
between
these
two
situations.
So
the
claim
made
in
[Thompson
&
Mann
86],
that
concession
is
a
matter
of
discourse
relations
rather
than
a
matter
of
the meaning
of
certain
conjunctions,
is
supported
by
the
ZPG
text.

45
8
Status
of
RST
RST
is
a
step
in
the
process
of developing
theories
of
communication.
Its
strengths
include
the
fact
that
it
provides comprehensive
analyses
rather
than
selective
commentary,
illustrated
in
its
assigning
a
purpose
and
status
to
every
unit
of
the
relational
elements
of
the
ZPG
letter.
Because
RST
is
applicable
to many
kinds
of
texts,
enabling
a
unified
description
of
text
structure
regardless
of
genre,
it
helps
to
factor
the
genuinely
genre-specific
aspects
of
text
from
the
more
genre-independent
ones.
There
are
several
aspects
of
text description
that
seem
particularly
important
for
communication,
but
which
it
does
not
yet address;
we
hope
that
these
will
be
addressed,
not
necessarily
by
us,
in
the near
future.
RST
has
not
yet
been
effectively
related
to
dialogue.
It
must
be
expanded
beyond
written
monologue
to
dialogue
and multilogue
in
order
to
encompass
a
fully
representative
range
of
the
functions
of language.
This
will
perhaps
be
the
most
significant modification
of
its
present
form.
RST
also
has
not yet
been
related to
all
of
the
kinds
of
theories with
which
it
should
interact.
Theories
of
holistic
structure
and
syntactic
structure
are
the
most
obvious
cases;
elaboration
in
both
directions
would
be
helpful.
In
these
cases
the
need
can
only
be
partly
met
by
augmentation
of
RST,
since one
really
needs
to
connect
to
theories
that
are
functional
in
the
same
effects-oriented
sense
that
RST
employs.
In
the
case
of holistic
structure,
there
are
function-oriented
accounts
that
suggest
that
a
useful
functional
reconciliation
is
possible.
Although
the
relationally-structured

46
parts
of
a
text
are
simply
parts
of
a
larger
whole,
they
are
the
parts
that
justify
and
motivate
the
existence
of
the
text. The
whole
is
thus
subordinate
to
the
part.
and
the
functions
of
the
nonrelational
parts,
such
as
the
signature
of
a
letter,
promote
the
effects
that
organize
the
relational
parts.
This functional
whole-under-part
organization
needs
investigation.
RST
makes
no
claims
about
the order
of
development
of
parts
of
a
text,
nor
about
how
that
order
might
limit
the
writer's
options.
Preliminary
studies
by Cecilia
Ford
and
colleagues
of
a
large
number
of
short texts indicated
that
neither
progressive
refinement
(top
down)
nor
aggregation
of
apriori
collections
(bottom
up)
was very
credible
as
a
simple
explanation
of
how
the
texts
might
have
been
developed.
A
complex mixed
strategy
seems
called
for;
more
study
is
needed.
A
systematic
description
of
how
relations
are realized
is
also
needed,
one
that
ranges
more
widely
than
any
study
of
conjunctions,
covering
the various
kinds
of
unsignalled
cases
as
well.
Beyond
these,
linking
RST to
theories
of
text
properties
such
as
information
flow,
thematic
structure,
and
lexical
relations
would
also
be
worthwhile.
9
Conclusion
We
have
presented
RST
as
a
way
to
account
for
the
functional
potential
of
text,
its
capacity
to achieve
the
purposes
of speakers
and
produce
effects
in
hearers.
The
principal
assumptions
that
RST
encodes
in
its methods
have
been
identified.
In
the
course
of
accounting
for
potential
effects,
we
have
also
shown
a
way to
distinguish
coherent
texts
from
incoherent
ones,
and identified
a
few
of
the
consequences
of
text
structure.

17
In
presenting
our
analysis
of
the
ZPG
letter
we
hope
that
the
met
hods
have
been
made explicit enough
so
that,
by
using
these
and
the
related
techniques
found
in
other
publications,
you
will
be
able
to
experiment
with RST and improve
it.

48
I.
Appendix:
Relation
Definitions
All
of
the
relations
used in
the
analysis, and
a
number
of
others,
have
been
defined
in
the
style described
above.
In
[Mann
&
Thompson
89],
all
of
the
definitions
are
given,
exemplified
and
discussed.
The definition
of
Evidence
was
discussed
at
length
in
Section
3.2
above.
In
this
section
we
present definitions
of
the
other
relations
used in
the
analysis,
with
only
occasional
brief
discussion.
In
the
definitions,
N
represents the
nucleus
and
S
the satellite.
Since
for
many readers
the
natural
examples
have
the
status
of
data
rather
than
just
exposition,
the
examples
used
in
[Mann
&
Thompson
89]
and
[Mann
&
Thompson
88]
to
illustrate
the relations
are
also
presented,
but without
unit
division
or
analysis.
We
should
note
that
the
relation
definitions
have
the
status
of
applications
of
the
theory
rather
than
elements
of
the theory.
One
might
want
to
change
or
replace
the
definitions
in
accounting
for
other
languages,
genres,
or
views
of
language
function
than
the
ones
that
led
to
these
definitions; such
changes
are
expected
and
do
not
cross
the
definitional
boundaries
of
RST.
The
relations
are
organized
as
follows:
Motivation,
Concession,
Background
Circumstance,
Solutionhood,
Elaboration,
Purpose,
Non-volitional
Result,
Means,
Restatement
Sequence,
Contrast,
Joint
The
first
group
has
only
the
nucleus
as
its
locus
of
effect,
the
second
group
has
nucleus
+
satellite,
and
the
third
group
is
not
nucleated.

49
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
pei-form
action
presented
in
N
the
effect:
R's
desire to
perform
action presented
in
N is
increased
locus
of
the
effect:N
From
a
personal
message
on
an
electronic
bulletin
board:
Text
Example:
"The
Los
Angeles
Chamber
Ballet (the
ballet
company
I'm
dancing with)
is
giving
4
concerts
next
week
...
Tickets
are
$7.50
except
for
the
opening
night
...
The
show
is
made
up
of
new
choreography
and
should
be
very
entertaining. I'm
in
3
pieces.
"
Several
relations
involve
notions
of
cause.
In
broadly
defining these
relations,
it
is
hard
to include
both
situations
that
are
intended
outcomes
of
some
action
and
causation
that
does
not
involve
intended
outcomes,
such
as
physical
causation.
Because
of
this difficulty,
we
have
divided
the relations
into
volitional
and
a
non-volitional
groups. Similarly
we
also
divide
them
on
the
basis
of
nuclearity
into
cause
and
result
groups.
Non-Volitional Result,
below,
is
one
of
the four
relations
so
produced.

50
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
the
compatibility
between
the
situations
presented
in
N
and
S
inc'eases
R's
positive
regard
for
the
situation
presented
in
N
the
effect:
R's
positive
regard
for
the
situation
presented
in
N is
increased
locus
of
the
effect:
N
and
S
Text
Example:
"Title:
Dioxin
Concern
that
this
material
is
harmful
to
health
or
the
environment
may
be
misplaced. Although
it is
toxic
to
certain
animals,
evidence
is
lacking
that
it
has
any
serious
long-term
effect
on
human
beings.
"
Text
Example:
"Although
Jim
lists
tenris,
Chinese
food,
and
travel
to
exotic
locales
among
his
favorite
hobbies,
one
can't
help
but
wonder
at
the
unmentioned
interests
that
help
spark
Jim's
creativity,
leading
him
to
concoct
an
unending
stream
of
imaginative
programs.'

51
Background
constraints
on
N:
R
won't
comprehend
N
sufficiently
before
reading
text
of
S
constraints
on
the
N +
S
combination:
S
increases
the
ability
of
R
to comprehend
an
element
in
N
the
effect:
R's
ability
to comprehend
N
increases
locus
of
the
effect:N
Text
Example:
'Home
addresses
and
telephone
numbers
of
public
employees
will
be
protected
from
public
disclosure under
a
new
bill approved
by
Gov.
George
Deukmejian.
Assembly
Bill
3100
amends
the
Government
Code,
which
required
that
the
public
records
of
all
state
and
local
agencies.
containing
home
addresses
and
telephone
numbers
of
staff,
be
open
to
public
inspection.
Circumstance
constraints
on
S:
S
presents
a
situation
(not
unrealized)
constraints
on
the
N +
S
combination:
S
sets
a
framework
in
the
subject
matter
within
which
R
is
intended
to
interpret
the
situation
presented
in
N
the
effect:
R
recognizes
that
the
situation
presented
in
S provides
the
framework
for
interpreting
N
locus
of
#hle
effect:
N
and
S
Text
Example:
"Probably
the
most
extreme
case
of
Visitors
Fever
I
have
ever
witnessed
was
a
few
summers
ago
when
I
visited relatives
in
the
Midwest."

52
Text
Example:
"P.
M.
has
been
with
KUSC
longer
than
any other
staff
member. While
attending
Occidental
College,
where
he
majored
in
philosophy,
he
volunteered
to work
at
the
station
as
a
classical
music
announcer.
That
was
in
1970."
Solutionhood
constraints
on
S:
presents
a
problem
constraints
or.
the
N +
S
combination:
the
situation
presented
in
N
is
a
(partial)
solution
to
the
problem
stated
in
S;
the
effect:
R
recognizes
the
situation
presented
in
N
as
a
(partial)
solution
to
the
problem
presented
in
S
locus
of
the effect:
N
and
S
Text
Example:
"One
difficulty
... is
with
sleeping
bags
in
which
down
and
feather
fillers
are
used
as
insulation.
This
insulation
has
a
tendency
to
slip
towards
the
bottom.
You
can
redistribute
the
filler.
...
In
the
definition
of
the
solutionhood
relation,
the terms
problem
and solution
are
broader
than
one
might
expect.
The
scope
of
problem includes:
1.
questions
2.
requests,
including
requests
for
information
3.
some
descriptions
of
desires,
goals,
intellectual
issues,
gaps
in
knowledge
or
other
expressions
of needs
4.
conditions
that
carry
negative
values,
either
expressly
or
culturally,
including
calamities
and
frustrations.
It
thus
compares
to Grimes'
Response
predicate
[Grimes
75].

53
Elaboration
constraints
on
the
N +
S
combination:
S
presents
additional
detail
about
the
situation
or
some
element
of
subject
matter
which
is
presented
in
N
or
inferentially
accessible
in
N in
one
or
more
of
the
ways
listed
below.
In
the
list,
if
N
presents
the
first
member
of
any
pair,
then
S
includes
the
second:
1.
set
:
member
2.
abstract
:
instance
3.
whole
:
part
4.
process
:
step
5.
object
:
attribute
6.
generalization
:
specific
the
effect:
R
recognizes
the
situation
presented
in
S
as
providing
additional
detail
for
N.
R
identifies
the
element
of
subject
matter
for
which
detail
is
provided.
locus
of
the
effect:
N
and
S
From
a
conference
announcement
brochure:
Text
Example: "Sanga-Saby-Kursgard,
Sweden,
will
be
the
site
of
the
1969
International
Conference
on
Computational
Linguistics,
September
1-4.
It
is
expected
that
some
250
linguists
will
attend
from
Asia.
West
Europe.
East
Europe
including
Russia,
and
the
United
States.
The
conference will
be
concerned
with
the
application
of
math
,tatical
and
computer
techniques
to
the
study
of
natural
languages,
the
development
of
computer
programs
as
tools
for
linguistic
research,
and
the
application
of
linguistics
to
the
development
of
man-machine communication
systems.

54
Purpose
constraints
on
N:
presents
an
activity
constraints
on
S:
presents
a
situation
that
is
unrealized
constraints
on
the
N +
S
combination:
S
presents
a
situation
to
be
realized
through
the
activity
in
N
the
effect:
R
recognizes
that
the
activity
in
N
is
initiated
in
order
to
realize
S
locus
of
the
effect:
N
and
S
Text
Example:
"To
see
which
Syncom
diskette
will
replace
the
ones
you're
using
now.
send
for
our
free
"Flexi-Finder"
selection
guide
and
the
name
of
the
supplier
nearest
you.
"
Text
Example: "Presumably,
there
is
a
competition
among
trees
in
certain
forest
environments
to
become
as
tall
as
possible
so
as
to
catch
as
much
of
the
sun
as
possible
for
photosynthesis."

55
Non-Volitional
Result
constraints
on
S:
presents
a
situation
that
is
not
a
volitional
action
constraints
on
the
N +
S
combination:
N
presents
a
situation
that
caused
the
situation
presented
in
S;
presentation
of
N
is
more
central to
W's
purposes
in
putting
forth
the
N-S
combination
than
is
the
presentation
of
S.
the
effect:
R
recognizes
that
the
situation
presented
in
N
could
have
caused
the
situation
presented
in
S
locus
of
the
effect:
N
and
S
Text
Example:
"The
blast, the
worst
industrial
accident
in
Al
exico's
history,
destroyed
the
plant
and
most
of
the
surrounding
suburbs.
Several
thousand
people
were
injured,
and
about
300
are
still
in
hospital.
Means
constrainta
on
N:
presents
an
action.
constraints
on
S:
none
conatrainta
on
the
N +
S
combination:
The
situation
presented
in
S
actually tends
to
make possible
or
likely
the
situation
presented
in
N.
the
effect:
R
recognizes
that
the
situation
presented
in
S
actually
tends
to
make
possible
or
likely
the
situation
presented
in
N.
locus
of
the
effect:
N
and
S

56
Text
Example:
"By
bouncing
sound
off
rock
layers
under
the sea
floor
and
recording
the
reflections
uzith
many
detectors,
structural
images
of
the crust
can
be
made
at
the
boundaries
where
plates
collide
and
rift
apart."
Restatement
constraints
on
the
N + S
combination:
S
restates
N,
where
S
and
N
are
of
comparable
bulk
the
effect:
R recognizes
S
as
a
restatement
of
N
locus
of
the
effect:
N
and
S
Text
Example:
'A
WELL-GROOMED CAR
REFLECTS
ITS OWNER
The
car
you
drive
says
a
lot
about
you.'
The last
three relations
--
Sequence,
Contrast
and
Joint
--
are
non-nucleated.
Sequence
constraints
on
N:
multi-nuclear
constraints
on
the
combination
of
nuclei:
A
succession
relationship
between
the
situations
is
presented
in
the
nuclei
13
the
effect:
R
recognizes
the
succession
relationships
among
the
nuclei.
locus
of
the
effect:
multiple
nuclei
Text
Example:
'Peel
oranges
and
slice
crosswise.
Arrange
in
a
bowl
and
sprinkle
with
rum
and
coconut.
Chill
until
ready
to
serve.
'
13
Note
that
the
definition
does
not
cover
presentational
sequence,
e.g.,
"First
... :
Second
...

57
Contrast
constraints
on
N:
multi-nuclear
constraints
on
the
combination
of
nuclei:
no
more
than
two
nuclei:
the
situations
presented
in
these
two
nuclei
are (a)
comprehended
as
the
same
in
many
respects
(b)
comprehended
as
differing
in
a
few
respects
and
(c)
compared
with
respect
to
one
or
more
of
these differences
the
effect:
R
recognizes
the
comparability
and
the
difference(s) yielded
by
the
comparison
is
being
made
locua
of
the
effect:
multiple
nuclei
Text
Example:
'Animals
heal,
but
trees compartmentalize.
They
endure
a
lifetime
of
injury
and
infection
by
setting
boundaries
that resist
the
spread
of
the
invading
microorganisms.
Joint
The
Joint
relation
is
a
notational
shorthand
for
the
JOINT
schema
defined
elsewhere.
Its
"relation
definition"
does
not
contain
any
constraints
or
specify
any
effects.
Text
Example:
'Employees
are
urged
to
complete
new
beneficiary
designation
forms
for
retirement
or
life
insurance
benefits
whenever
there
is
a
change
in
marital
or
family
status.

58
Employees
who
are not
sure
of
who
is
listed
as
their
beneficiary
should
complete
new
forms
since
the
retirement
systein
and
the
insurance
carrier
-use
the
most
current
form
to
disburse
benefits.'

59
References
[Cui
85]
Cui,
Songren,
Comparing
Structures
of
Essays
in
Chinese
and
English.
Master's
thesis,
UCLA,
1985.
[Ford
871
Ford.
Cecilia.
"Overlapping Relations
in
Text
Structure."
in
Scott
DeLancey
&
Russell
Tomlin
(eds.),
Proceedings
of
the
Second
Annual
Meeting
of
the
Pacific
Linguistics
Society,
Linguistics
Department,
University
of
Oregon.
1987.
[Fox
871
Fox,
Barbara
A..
Discourse Structure and
Anaphora
in
Written
and
Conversational
English,
Cambridge University Press,
1987.
[Grimes
751
Grimes,
J.
E.,
The
Thread
of
Discourse,
Mouton.
The
Hague,
1975.
[Halliday
85]
Halliday,
M.A.K.,
An
Introduction
to
Functional
Grammar,
Edward
Arnold,
London,
1985.
[Halliday
&
Hasan
88]
Halliday,
M.
A. K.
and
Ruqaiya
Hasan.
Language.
Context
and
Text:
a
social
semiotic
perspective,
Oxford
University Press,
1988.
[Hasan
781
R.
Hasan,
"Text
in
the Systemic-Functional
Model."
in
W.
Dressier
(ed.).
Current
Trends
in
Text
Linguistics,
de
Gruyter,
1978.
[Kumpf
86]
Kumpf,
Lorraine,
Structuring Narratives
in
a
Second
Language:
a
Description
of
Rhetoric
and
Grammar,
Ph.D.
thesis.
University
of
California
at
Los
Angeles,
1986.
[Larkin
&
O'Malley
73]
Larkin,
Don
and
Michael
H.
O'Malley,
"Declarative
Sentences
and
the
Rule-Of-Conversation
Hypothesis,"
in
Papers from.
the
Ninth
Regional
Mfeeting,
pp.
306-319,
Chicago Linguistic
Society,
April
1973.
[Mann
90]
Mann,
William
C.,
The Origins
of
Text
Structuring
Relations,
1090.
In
Preparation.
[Mann
&
Thompson
86]
Mann, William
C.
and
Sandra
A.
Thomps,'n.
"Relational
Propositions
in
Discourse,"
Discourse
Processes
9,
(1),
.January-March
1986.
57-90.
Also
available
as
ISI/RR-83-115.
[Mann
&
Thompson
d7a]
Mann,
William
C.
and
Thompson.
Sandra
A..
Rhetorical
Structure
Theory:
A
Theory
of
Text
Organization.
USC
Information
'ciences
Institute,
Technical
Report
ISI/RS-87-
190, 1987.
[Mann
&
Thompson
87b
Mann,
William
C.
and
Sandra
A.
Thompson.
"Rhetoric'fl
Structure
Theory:
A
Framework
for
the Analysis
of
Texts."
IPRA
Papers
in
Pragmatics
1,
1987,
79-105.
Also
available
as
report lSI/RS-87-185.
[Mann
&
Thompson
88]
Mann,
William
C.
and
Thompson.
Sandra
A..
"Rhetorical
Structure
Theory:
Toward
a
Functional
Theory
of
Text
Organization."
Text
8. (3).
1988.
213-281.
This
is
an
abridged
version
of
[Mann
&"
Thompson
89].

60
[Mann
&
Thompson
89]
Mann,
William
C.
and Thompson,
Sandra
A.,
"Rhetorical
Structure
Theory:
A
Theory
of
Text
Organization,"
in
Livia
Polanyi
(ed.),
The
Structure
of
Discourse,
Ablex,
Norwood,
N.J.,
1989.
Also
available
as
ISI/RS-87-190
from
Document
Center, USC/ISI,
4676
Admiralty
Way.
Marina
del
Rey,
CA
90292.
[Mann
&
Thompson
901
Mann,
William
C.
and
Sandra
A.
Thompson,
Discourse
Description:
Diverse
Analyses
of
a
Fund
Raising
Text,
John Benjamins,
Amsterdam,
1990.
Volume
in
preparation.
[Matthiessen
&
Thompson
89]
Matthiessen,
Christian
and
Thompson,
Sandra
A.,
"The
Structure
of
Discourse
and
"Subordination","
in
Haiman
and
Thompson
(eds.),
Clause
Combining
in
Grammar
and
Discourse,
Benjamins,
Amsterdam,
1989.
Also
available
as
ISI/RS-87-183.
[Noel
86]
Noel,
Dirk,
Towards
a
Functional
Characterization
of
the
News
of
the
BBC
World
News
Service.
Antwerp,
Belgium,
1986.
Antwerp
Papers
in
Linguistics,
Number
49.
[Reddy
79]
Reddy,
Michael,
"The
Conduit
Metaphor,"
in
A.
Ortony
(ed.),
Metaphor
and
Thought,
Cambridge
University Press,
Cambridge,
1979.
[Thompson
&
Mann
86]
Thompson,
Sandra
A.
and
William
C.
Mann,
"A
Discourse
View
of
Concession
in
Written
English,"
in
Scott
DeLancey
&
Russell
Tomlin
(eds.),
Proceedings
of
the
Second
Annual
Meeting
of
the
Pacific
Linguistics
Conference,
Pacific
Linguistics
Conference,
November
1986.
[Thompson
&
Mann
87]
Thompson,
Sandra
A.
and
William
C.
Mann.
"Antithesis:
A
Study
in
Clause
Combining
and
Discourse
Structure
,"
in
Ross
Steele
and
Terry
Threadgold
(eds.),
Language
Topics:
Essays
in
Honour
of
M.
A.
K.
Halliday,
Benjamins, Amsterdam,
1987.
Also
available
as
ISI/RS-87-171.
[van
Dijk
72]
van Dijk,
Teun
A.,
Some
aspects
of
text
grammars,
Mouton,
The
Hague,
1972.
[van
Dijk
771
van Dijk,
T.,
Text
and
Context:
Explorations
in
the
Semantics
and
Pragmatics
of
Discourse,
Longman,
1977.
[van
Dijk
80]
van Dijk,
Teun
A..
Macrostructures,
Erlbaum,
Hillsdale,
NJ,
1980.
[van
Dijk
81]
van
Dijk,
Teun
A.,
Studies
in
the
pragmatics
of
discourse,
Mouton.
The
Hague,
1981.
Ivan
Dijk
85]
van Dijk,
Teun
A.,
Semantic
Discourse
Analysis,
Academic
Press,
pp.
103-136,
1985.