STRUCTURES AND MEANINGS

A WAY TO INTRODUCE
ARGUMENTATION ANALYSIS IN POLICY STUDIES EDUCATION

Des Gasper

May 2000

Working Paper 317

0
The Institute of Social Studies is Europe's longest-established centre of higher education and
research in development studies. Post-graduate teaching programmes range from six-week
diploma courses to the PhD programme. Research at ISS is fundamental in the sense of laying a
scientific basis for the formulation of appropriate development policies. The academic work of
ISS is disseminated in the form of books, journal articles, teaching texts, monographs and
working papers. The Working Paper series provides a forum for work in progress which seeks
to elicit comments and generate discussion. The series includes the research of staff, PhD
participants and visiting fellows, and outstanding research papers by graduate students.

For further information contact:

ORPAS - Institute of Social Studies - P.O. Box 29776
2502LT The Hague - The Netherlands - FAX: +31 70 4260799
E-mail: workingpapers@iss.nl

ISSN 0921-0210

Comments are welcome and should be addressed to the author:

ABSTRACT
The well-known Toulmin-Dunn model for considering structure in policy argu-
ments has to be subordinated to a more general method, such as Scriven’s, for case-by-
case examination of texts: to reflect on meanings, flexibly identify the linkages between
elements, and carefully draw out unstated conclusions and assumptions. The paper pre-
sents some simple supplementary methods to help do this: a tabular format for exam-
ining the components of an argument and their meanings; second, attention to the use of
terms that convey praise or criticism and hence hint at conclusions but also, as one be-
comes conscious of the linguistic choices made, point towards possible counter-
arguments; and thirdly, a simpler tabular version of the Toulmin-Dunn categories, to
describe an argument’s logical structure and the possible rebuttals. Detailed worked ex-
amples analyse speeches of the Zimbabwe Minister of Local Government, a diagnosis
of the phases of local government formation in independent India, and a South African
editorial on population growth.

1
 1
CONTENTS
1. Introduction ..................................................................................................................3
2. Contributions and dangers of the Toulmin-Dunn format for analysing policy
arguments .....................................................................................................................4
Figure 1: A version of the Toulmin-Dunn model for policy arguments.....................6
Figure 2: Failings in use of the Toulmin-Dunn format; and some relevant
responses.....................................................................................................8
3. A more flexible and balanced approach: guiding Toulmin-Dunn BY Scriven’s seven
steps of argument analysis ...........................................................................................8
Figure 3: Scriven’s seven-step model for argument analysis .....................................9
4. Centralization is decentralization ? Minister Chikowore on supervision of local
government in Zimbabwe..........................................................................................11
Figure 4: A step-wise analysis of Minister Chikowore’s defence of his powers .....13
Figure 5: The structure of Minister Chikowore’s two claims about
(de)centralization......................................................................................14
Figure 6: Examples of praise terms and corresponding criticism terms...................15
5. Veiled conclusions: Minister Chikowore on NGOs in Zimbabwe............................16
Figure 7: Step-wise analysis and paraphrase of Minister Chikowore’s comments on
NGOs........................................................................................................17
Figure 8: The logical structure of Minister Chikowore’s comments on NGOs .......18
6. Mixed and Qualified Messages: Harry Blair on Rural Local Government in India..19
Figure 9: Step-wise analysis and paraphrase of Blair’s description of historical
stages ........................................................................................................20
Figure 10: Analysis of the structures of Blair’s suggestions .....................................21
7. A more extended argument: ‘the Star’ on population growth in South Africa .........22
Figure 11: An informal structure diagram of the ‘China Syndrome’ editorial.........25
Figure 12: The argumentative structure of ‘The China Syndrome’’s main claims ..26
8. Conclusion..................................................................................................................26
References ......................................................................................................................27
Endnotes .........................................................................................................................27

2
1. INTRODUCTION
There are many aspects of ‘doing policy things with words'. I focus here on
policy wording as argumentation, where policy talk and writing are supposed to make
logically reasoned claims. Policy analysis centrally involves the assessment and prepa-
ration of arguments in which ideas about values/objectives/priorities are combined with
claims about facts and cause-effect linkages, to produce valuations about past or possi-
ble future actions by agents operating in the public arena. Each approach in policy
analysis can be seen as a particular style of building arguments, which selects and han-
dles ideas and data in its own distinctive way.
In presenting policy analysis as argumentation I want to avoid excessive con-
trasts between argument analysis and (other) discourse analysis, and between text-
focused and context-focused types of analysis.1 The structural approach in argument
analysis only proves fruitful when combined with a careful attention to words and
meanings, and these depend on contexts. When students, and practitioners and aca-
demics, of policy and planning use standard structure formats to analyse its texts, but
without a primary focus on meanings and messages, the result is often confusion. The
standard formats such as Toulmin-Dunn’s or the ‘logical framework’ can then quieten
fears but breed thoughtless errors (Gasper & George 1998; Gasper 1997, 2000). Simi-
larly, argument evaluation conducted by reference to a standard list of fallacies in infer-
ence (such as in Thouless 1974, or Sillince 1986) can go astray if users seek to imitate
and reproduce the examples without close attention to the meanings and context of their
particular text.
We need an approach that gives balanced attention to elucidation of meanings,
analysis of structure, and evaluation of cogency. I propose Scriven’s format for argu-
ment analysis as helpful here. I further suggest how to operationalize its search for
meanings in a fashion accessible to ordinary students of policy analysis, through an
analytic table plus attention in particular to language of praise and condemnation; and
how to link this to a user-friendly tabular version of the Toulmin-Dunn format, to or-
ganize ideas about argument components into a picture of argument structure.
The paper then illustrates how these tools – post-Scriven tables to investigate
meanings and a post-Dunn table to record structure -- can provide insights into a series

3
of policy-related texts, mainly from Southern Africa. We will look at four texts. First, a
relatively straightforward defence of a controversial policy by a government minister --
which yet on examination proves more complex than it first appears. Second, a more
elusive statement by the minister, in which all the conclusions are hinted at, conditional,
not explicit and definite. Third, another elusive, allusive text, an attempt to describe and
interpret enormous, multi-faceted, unpredictable historical movements. Finally, a longer
text, to show how the tools of close examination remain workable and helpful at a
greater scale, given some degree of selectivity.

2. CONTRIBUTIONS AND DANGERS OF THE TOULMIN-DUNN

FORMAT FOR ANALYSING POLICY ARGUMENTS
Policy reasoning involves far more extended argumentation than the brief syllo-
gisms in logic texts or even than utilitarianism's identification and summation of ef-
fects. Which effects are relevant, in terms of which values and responsibilities, and
judged by which comparisons? Which rights and duties must be respected, besides fu-
ture consequences? Which constraints have been considered and how? Which issues
demand attention in the first place? How are conclusions to be drawn from varied and
typically incomplete information? Tools of argumentation complement the standard
emphases in policy analysis and planning on measurement and calculation. They are
vital when there is no time or capacity for new empirical investigations, and important
in every case.
Teaching and learning about any type of argumentation face the complexities of
trying to identify the structures from the meanings and understand the meanings from
the structures. Besides the clarification of terms and of how words combine in single
sentences, we have to look at how single propositions link up in broader discourses.
Teaching and learning about argumentation in a specialist area like policy discourse
face further issues. Policy analysts and planners lack the time, background and interest
to immerse themselves in complex approaches from logic or linguistics. Formal logic
has anyway not been concerned with ongoing practical and policy debates and is instead
a fundamental discipline with its own concerns. Nor is discourse theory yet very helpful
here; it requires a considerable investment of learning but fails to go into the specifics

4
of public policy. We need some serviceable approaches that even though imperfect will
allow average analysts to do better when they consider and prepare arguments. Thus
more relevant and less abstruse are some studies in the past generation oriented to prac-
tical argumentation (e.g. Scriven 1976). In America they are called "informal logic" and
are part of the work on "critical thinking". Some authors have gone on to apply this to
understanding policy and planning arguments (Gasper 1996a gives a survey).
The Toulmin model of argumentation, created by the British-American philoso-
pher Stephen Toulmin (1958, 1979) has had great influence in a variety of specialist
fields outside philosophy, as one such serviceable approach. It is intended to apply to
any argument, and contains nothing specific to policy arguments. However, a large pro-
portion of the attempts to formalize argumentation analysis in planning and policy
analysis have used his model, especially as adapted by the University of Pittsburgh
policy theorist William Dunn (1981, 1994). It highlights how policy arguments have
structures, and the range of types of justification used.
In the classical syllogism a conclusion follows inexorably from the combination
of premises: typically a premise that states a contingent circumstance and another that
states a general principle. Toulmin expresses these three central elements in an argu-
ment as follows:
• the Claim, the proposed conclusion. This label captures that a Claim is debatable;
hence the need for the argument.
• Grounds (Toulmin et al. 1979; called Data in Toulmin 1958), which are presented
as supporting the Claim; there may be one or more;
• the Warrant. If the truth of the Claim does not automatically follow from the
Ground(s), then a linking statement, a warrant for the inference, must be given.
Typically the Warrant is seen as a rule or principle relevant to a number of cases.

Toulmin's schema then adds some categories needed for practical argumentation but not
found in the classical syllogism: layers of back up, non-definitive inferences, exceptions
and counter arguments.
• The Warrant itself may be debatable and require justification by a Backing state-
ment or statements.

5
• Often a Qualifier is used to modulate the Claim: a word like ‘probably’ or ‘appar-
ently’, or a marker of exceptions like ‘unless’. For not only can Warrants be debat-
able in terms of relevance or their own validity, other counterconsiderations may
exist.
• A Rebuttal is such a counterconsideration. It may concern any of the elements of the
argument; for example it may directly concern the Claim. It may itself require
Backing.

Figure 1 indicates the relationships of these components within a policy argu-

ment, largely following Dunn’s adaptation of Toulmin. A horizontal chain runs from
Grounds to the Claim, via an inference (‘So’/'Therefore’) and a Qualifier. In one verti-
cal chain, this inference is supported by a Warrant, supported in turn by Backing. In an-
other vertical chain, the Qualifier is supported by a Rebuttal which could weaken the
force of the Claim.

Figure 1: A version of the Toulmin-Dunn model for policy arguments

Given
GROUNDS CONCLUSION
(policy-relevant (policy proposal
information) Therefore QUALIFIER / CLAIM)
(inference) (modulates strength of
the conclusion)
And Since
WARRANT Unless
(justification for the inference) REBUTTAL

Because Because
BACKING BACKING

While both Rebuttal and Warrant could themselves often be seen as Claims,
resting on Grounds, a Warrant and so on, in the Toulmin-Dunn diagram this is only re-
ferred to by use of Backing elements for them. This is to retain a clear focus on a par-
ticular Claim, while recognising the presence typically of a system of arguments more
complex than the simple syllogism. Similarly, though Rebuttals may concern any ele-
ment of the argument, Fig. 1 visually links Rebuttals only to the categories of Qualifier
and Claim, rather than complicate the diagram and risk loss of focus. The central Claim

6
must remain highlighted, because it is only by reference to this that the other elements
can be identified: they derive their roles in relation to it.
The Toulmin-Dunn format helps us to see that:
• arguments have functionally distinct components (e.g. warrants);
• arguments are diverse (e.g. there are many types of warrant; Dunn 1994, Ch.4);
• arguments have structures, including, often, layers (e.g. warrants may need back-
ing);
• most arguments are not certain; they concern ‘open systems', and are vulnerable to
counter-arguments (potential ‘rebuttals'); hence the need for ‘qualifiers';
• the same data/grounds can be interpreted in different ways, depending on the war-
rants and other propositions with which it is combined; argumentation is creative,
interpretive (Dunn 1994).2

Limitations of the Toulmin-Dunn model, and some responses

While the format serves to make several important general points, it cannot be
used as a template to specify the shape of every argument that one encounters. Unfortu-
nately it is often handled as if it could. Gasper & George (1998) note that not only do
beginner students tend to misuse the format as a template, so do several published
pieces on policy argumentation (e.g. Bozeman & Landsbergen 1989; Ray 1990; even
Dunn 1994).
• Frequently, complex arguments that involve multiple steps and require multiple
diagrams are squeezed into a single diagram copied from the introductory state-
ments and examples given by Dunn or Toulmin.
• The classification (‘coding’) of argument components in terms of Toulmin’s catego-
ries becomes very erratic when the contours of the particular argument being con-
sidered are overshadowed by the authority of those exemplars.
• The Toulmin diagram appears to cow many users, even some with computer skills,
who find it difficult to draw and re-draw, and hence seem content to imitate a stan-
dard layout regardless of the specific content of the argument concerned. Further,
the Claim’s location at the top-right does not promote its identification as the first
step; yet correct identification of all other components depends on that step.

7
 Figure 2 summarizes Gasper & George’s findings and corresponding sugges-
tions. In this paper I take up in particular the following ideas:
• to break texts into multiple arguments, which requires a layered presentation;
• to use a tabular presentation, rather than the Toulmin-Dunn (T-D) diagram; and
• to prepare that presentation by use of an argument specification procedure that con-
centrates on meanings, not on pre-specified labels or a standardized structure dia-
gram.

Figure 2: Failings in use of the Toulmin-Dunn format; and some relevant re-
sponses
COMMON WEAKNESSES INTERNAL RESPONSE EXTERNAL RESPONSE
IN USE OF T-D FORMAT (staying inside T-D format) (going beyond T-D format)
1. Oversimplification, e.g. by Multiple diagrams E.g. Add a picture of multiple
sticking to single diagram and levels of discourse (e.g. Fischer-
thus excluding too much Taylor format)

2. Slow and unreliable ‘coding’ of Provide procedural advice on Represent arguments without first
components (i.e. classification as ‘coding’ labelling components (e.g. as in
claim/data/warrant/backing…) Scriven’s approach)

3. Use of same diagrammatic Vary the shape of the diagram, Adopt/teach an argument
layout for all cases according to content of the case specification procedure (e.g.
Scriven’s), not a standard layout

4. Difficulty in using diagram, as it Use a system of columns; and/or Concentrate on other methods of
does not highlight the starting master computer graphics tools. argument specification
point for analysis and if it is
found hard to modify

3. A MORE FLEXIBLE AND BALANCED APPROACH: GUIDING

TOULMIN-DUNN BY SCRIVEN’S SEVEN STEPS OF ARGUMENT
ANALYSIS
The Toulmin/Dunn format is only an introduction. We need also, first, more
complex formats that highlight the specifics of policy--the characteristic nature of war-
rants, backings, rebuttals, and qualifiers in policy arguments. Hambrick (1974) pro-
vided a useful listing of standard components of a well-designed policy argument,
which can be grouped into three stages (Gasper 1996a). The first stage, elaboration of
cause-effect links (similar to problem analysis in the logical framework approach),
stresses the background role of world-views ("grounding propositions"). It seeks to es-

8
tablish that action on a particular policy instrument will lead to increase of another vari-
able taken as a goal. Hambrick’s second stage covers normative inputs to the argument,
to turn cause-and-effect into means-and-end: what justifies the specified goal? why is
something considered a problem and its reverse a desirable end? why are certain types
of action considered acceptable? This type of normative analysis is elaborated in
Fischer's model, which looks at the basis of the normative and grounding propositions.
The third stage that one can discern in Hambrick contains necessary tests of a policy
argument, i.e. of a claim that specified actions will lead to desirable ends and should be
done. It includes especially questions about constraints and alternatives.
All such formats have dangers if seen as a universally valid template into which
any argument can be forced. So they must, secondly, be complemented by a flexible
and exploratory approach, such as that of Michael Scriven (1976). Scriven highlights
how both identifying an argument's content and structure, and then assessing them, are
complex activities and require systematic and separate handling. He allows for the like-
lihood that the structure (the system of links between components, and their respective
roles). will differ from case to case. Figure 3 outlines his model of good argument
analysis, which contains seven distinct steps. Almost any step can lead back to earlier
ones.
Figure 3: Scriven’s seven-step model for argument analysis
ARGUMENT SPECIFICATION ARGUMENT EVALUATION
1. Clarify meanings [of terms] 5. Criticize inferences and premises

2. Identify conclusions, stated and un- 6. Consider other relevant arguments [in-
stated cluding possible counter-arguments]

3. Portray structure [the relation of conclu- 7. Overall evaluation [based on the bal-
sions to grounds and warrants] ance of strengths and weaknesses, and
in comparison to the alternative
4. Formulate unstated assumptions [i.e. stance(s) one could instead adopt]
those required to move from the
grounds to the claims]

• Scriven presents identification of components as part of step 3, portrayal of struc-

ture. In my experience an initial identification of component parts is a necessary
preliminary--step 0, important for focussing attention, deciding which terms require

9
 semantic investigation, and asking about their roles. It may be revised later in step
3, when the system of roles is posited.
• In Section 4 below we consider one simple tool in step 1’s clarification of meanings
of terms: identification of praise-language and criticism-language, which provides
hints about unstated conclusions (step 2). In Sections 5 and 6 I illustrate how para-
phrasing a text into language with a different slant (praise/neutrality/criticism) can
also help in this.
• Toulmin's model concerns Scriven’s steps 2 to 4, identification of premises and infer-
ences. That work is a prerequisite for argument evaluation. In addition the attention to
possible rebuttals connects to step 6, consideration of other relevant arguments.
• Within argument evaluation, step 5 includes checking the strength of both the logic
and the assumptions/premises, stated and unstated. For while an argument can al-
ways be made logical by adding assumptions that would make its parts correctly
link, if those assumptions fail then the argument fails.
• Step 6 checks on possible counterarguments. This is again a way of checking logic.
Often we can restate a counterargument as a required but unrealistic unstated as-
sumption. Paraphrasing into language with a different slant helps in this identifica-
tion of counterarguments too.

In this paper I concentrate on Scriven’s steps 1 to 4, argument specification, the

necessary prelude to argument evaluation; but with attention also to step 6 on possible
counterarguments.
I suggest that a three-fold combination can be effective:
• Scriven’s procedure of step-wise analysis. It gives a framework for investigating
meanings and structure without preconceived and possibly distorting notions about
what are the components in a given text and their roles and linkages.
• The identification of praise- and criticism-language, for use in Scriven’s step 1, as
one way to focus attention on meanings.
• George’s tabular version of the Toulmin-Dunn format for summarizing argument
structure (step 3). The T-D format gives one way to tidily present the results from
the necessary careful, open-minded examination of possible components, roles and

10
 linkages. The standard option for portraying structure is a diagram, but an easier and
effective option is to organize the components identified in steps 1 and 2, including
the unstated conclusions, into a set of columns: RV George’s alternative format.
The columns lead to a clear grouping of components. This in turn can help the step
4 activity of identifying unstated but required assumptions and can thus enrich the
picture of an argument’s structure.

The next section illustrates this three-fold combination with an example from
Zimbabwe. One could ask students to do their own analysis of the given text before
studying the analysis below.

4. CENTRALIZATION IS DECENTRALIZATION ? MINISTER

CHIKOWORE ON SUPERVISION OF LOCAL GOVERNMENT IN
ZIMBABWE
Speaking at the 1989 Urban Councils Association meeting, the Zimbabwean
Minister of Local Government defended his acquisition and exercise of sweeping pow-
ers to intervene in local Councils. The Minister rejected
"claims at your previous conferences and other fora that Central Government
have tended to centralise power instead of decentralising... much of your
contention arises from the 1986 amendment to the Urban Councils Act which
gives the minister the power to approve the appointment and discharge of senior
officials... This was done... [after] serious problems regarding staff harassment
by mayors or councillors who were...wanting to get rid of officials whose faces
they dislike or those who do not come from where they come from" (Chikowore
1989b: 3-4).
To examine this text:
• We first divide it into components (step 0): in this case the three sentences, each of
which has a distinct message. Possibly the last sentence could be divided into two,
but its second part provides an explanation and definition of the first part’s allega-
tion of harassment.
• We next look word-by-word at each component. This allows us to see aspects we
could otherwise easily overlook. Here we find that the Minister has acquired powers
over both appointment and discharge, though he only gives as explanation a point
concerning discharges. It also leads us to identify and reflect on the choice of some

11
 key terms (step 1). In particular the label ‘harassment’ is a central choice, a descrip-
tion which is at the same time a condemnation.
• We can thus impute an unstated conclusion (step 2): that the Minister was right to
acquire such sweeping powers.
• We can then further impute two unstated assumptions (step 4): that the Minister
will do better in appointments than the Councils, and that there will not be major
undesirable side-effects such as demoralization of local government. The inverse of
those assumptions could alternatively be stated as proposed rebuttals to the Minis-
ter’s position.

A table of the type used in Figure 4 helps to ensure that we do not skip steps or
segments of text. Even if we carry out the examination initially without it, the table
helps us to later check and clearly present our analysis.

12
 Figure 4: A step-wise analysis of Minister Chikowore’s defence of his powers
Step 0. Identify components Step 1. Clarify meanings of Step 2. Identify conclusions
terms Step 4. Identify assumptions
a) [I reject] claims at your ‘claims’ can be an anti-term, Half-stated conclusion: Cen-
previous conferences and compared to ‘proposals’, tral Government has decen-
other fora that Central ‘contentions’, ‘statements’, tralised power
Government have tended to etc.
centralise power instead of
decentralising

b) ...much of your contention A neutral descriptive clause

arises from the 1986
amendment to the Urban
Councils Act which gives
the minister the power to NB: Both appointment and
approve the appointment discharge
and discharge of senior
officials...

c) This was done... [after] se- ‘harassment’ is typically an [Unstated assumption: Over-
rious problems regarding unfavourable term, compared riding powers for the Minister
staff harassment by mayors to e.g. ‘demanding efficiency’ do not constitute harassment
or councillors who of mayors and councillors.]
were...wanting to get rid of ‘whose faces they dislike’ is [Unstated assumption: staffing
officials whose faces they an unfavourable term sug- must not be on ascriptive or
dislike or those who do not gesting petty personal con- affective grounds.]
come where they come flicts
from.

Nothing is said about prob- [Unstated assumptions: may-

lematic new appointments ors and councillors brought
ascriptive and affective crite-
ria also into appointments -
this is why they sought to get
rid of other staff and why the
centre took powers over new
appointments too.]

[Unstated conclusion: gov-

ernment was right to pass the
amendment]
[Unstated assumptions:
- central government will do
better in staffing decisions
than would local politicians
- there are no important nega-
tive effects like loss of legiti-
macy and motivation of local
politicians.]

13
 Some of the unstated assumptions listed in Figure 4 might only emerge from
doing a structure analysis like the one in Figure 5. It corresponds to Scriven’s step 3 (a
sketch of structure) but also goes further to consider possible counterarguments. Since
we found two implied conclusions in the text we require two argument structures. Fig-
ure 5 gives attention first to the one which probably represents the major focus, defence
of the 1986 amendment.
The re-formatting of the Toulmin model in terms of columns allows a very clear
presentation, that matches the rhythms of everyday language and also reflects Scriven’s
logical sequence of analysis: first look for the Claim, then for the proposed supporting
Grounds (Data), then for the proposed principle that is supposed to bridge the move
from Grounds to Claim, and then for possible counter-arguments. Figure 5 uses four
columns but one could use more, e.g. also have a column for backing of warrants.

Figure 5: The structure of Minister Chikowore’s two claims about

(de)centralization
I PROPOSE THAT GIVEN THAT [DATA] AND THE UNLESS
[CLAIM], PRINCIPLE THAT [REBUTTAL]
There were serious [WARRANT],
[Unstated conclusions] problems regarding staff [Unstated possible
harassment by mayors or [Unstated assumption:] counter-arguments:]
Government was right to councillors who were Staffing must not be on - Central government
pass the amendment, wanting to get rid of ascriptive or affective will not do better in
acquiring a veto over officials whose faces grounds staffing decisions than
appts. and dismissals they dislike or those who would local politi-
do not come where they cians.
come from; [Unstated assumption:] - There are important
and [unstated Overriding powers for negative effects like
assumptions] mayors and the minister do not loss of the legitimacy
councillors brought as- constitute harassment of and motivation of
criptive and affective mayors and councillors local politicians
criteria also into (or are excusable)
appointments

[Half-stated conclu- and implied unstated arguments for it]

sion:]
the measure seeks to good centralisation is not
Central Government has counter a weakness in ‘really’ centralisation,
decentralised power, decentralization (namely and bad de-centralisation
that Councillors and is not ‘real’
even though [potential mayors harass staff), decentralisation; so this
rebuttal:] legislation is not anti-
it has centralised power; decentralisation

14
 Teaching of policy argumentation seems to have concentrated on description of
structure, such as by a Toulmin-Dunn diagram or logical framework, and assessment of
logic, aided for example by a warning list of fallacies. These are steps 3 and 5 in
Scriven’s model. Unless matched by careful handling of the earlier steps, especially
step 1’s reflection on meanings, the structure analysis and logic check are liable to miss
the point. A key part of our analysis of Minister Chikowore’s statement on local gov-
ernment was identification of terms that convey praise or criticism, leading to sugges-
tions of some implied or hinted conclusions. This provides a workable starting point for
students. One can ask them to identify partners for praise-terms or criticism-terms, such
as listed in Figure 6.

Figure 6: Examples of praise terms and corresponding criticism terms

PRAISE TERMS CRITICISM TERMS
Coordination Bureaucracy
Participation Anarchy
Democracy Populism; dissidents, deviants
Decentralization Disintegration
Full coordination Centralization

These examples come from a study of the discourses of decentralization in Zim-

babwe in the 1980s (Gasper 1991, 1997). Some other terms seen there, while not quite
praise, did sweeten messages. The exotic labels ‘chef’ and ‘povo’ dignified the posi-
tions of the new bosses and the bossed, those who spoke and those who listened (Gas-
per 1997: 52).
Examination of praise-terms and criticism-terms can lead on to broader exami-
nation of normative versus neutral language. In development discourse the most basic
term to examine is ‘development’. It is a normative praise term; yet is typically also
treated as neutrally descriptive. (The same applies to ‘efficiency’, ‘decentralization’, or
the other praise terms mentioned above.) Unless we try to distinguish these two aspects
in our minds, our normative choices become veiled and confused; and particular policy
strategies which become defined as ‘development’ can then be immune to criticism,
able to dismiss real cases of failure as not ‘real’ cases of themselves (see e.g. Apthorpe
& Gasper 1982, Gasper 1996b).

15
 Methods for investigating meanings form a huge topic. For purposes of sharp-
ening the appreciation of policy students, the two devices illustrated above are a
worthwhile start: a column for line-by-line and where necessary word-by-word reflec-
tion on meanings; and identification of praise- and criticism-terms.3
Our next extended example goes further into the meanings of terms and phrases
than has our look at Minister Chikowore’s defence of his powers over local govern-
ments. This time we meet the Minister when he turned his attention to NGOs.

5. VEILED CONCLUSIONS: MINISTER CHIKOWORE ON NGOS IN

ZIMBABWE
In a 1986 speech the Minister of Local Government in Zimbabwe declared that,
while "we all want the work of NGOs to increase", some NGOs ignored the local gov-
ernment machinery.
"[They act] as if they were unguided missiles... could it be that they have sinister
motives to distort our development process? Could it be that their actions are an
effort to impose a different development philosophy other than that chosen by
our people through their party and government? Or could it be that they just do
not wish to work in harmony with government?" (Chikowore 1986).

Figure 7’s analytic table differs from the one used earlier. The first column still
consists of a classification of the text into component sections. To leave space for an
exercise in alternative wordings, as in the third column, all comments that correspond
to Scriven’s steps 1 through 4 (on meanings, roles, implications and assumptions) have
been placed in the middle column. The third column then builds from the middle col-
umn’s comments on praise terms and criticism terms, to note how the behaviour those
terms refer to could be described with a different tone. This exercise gives contrasts
which help us to identify unstated conclusions and thus the overall meaning and struc-
ture of the argument-system, and also possible counter-arguments.

16
 Figure 7: Step-wise analysis and paraphrase of Minister Chikowore’s comments
on NGOs
THE TEXT, COMMENTS, IN MORE NEUTRAL OR
Divided into sections on meanings and logic FAVOURABLE LANGUAGE
".. we all want the work of NGOs Unstated conclusion: so you can- Some NGOs act as if they are not
to increase.. not say I am anti-NGOs branches of government but in-
stead have independent sources
[But some NGOs act] as if they ‘unguided missiles’ is 1. an anal- of legitimacy.
were unguided missiles… ogy, 2. Unfavourable
Do they wish to influence the
…could it be that they have sin- ‘sinister motives’ is unfavour- development process?
ister motives to distort our de- able;
velopment process? ‘distort’ is also unfavourable; Are they advocating and under-
‘our’ suggests there are outsiders taking actions not identical with
Could it be that their actions are interfering those advocated by the present
an effort to impose a different government ? - actions which are
development philosophy[,..] ‘ímpose’ is unfavourable chosen by the party leaders, for-
other than that chosen by our mally accepted by the party, and
people through their party and ‘chosen’ is favourable; so is ‘our hardly known by our people.
government? people’ and therefore so too is
‘their party’ They do not automatically agree
Or could it be that they just do [Unstated assumption: our peo- with what government proposes
not wish to work in harmony ple have only one party]
with government?” (Chikowore
1986) ‘just do not wish to’ suggests
closed minds; ‘work in harmony’
is favourable, so rejecting it is
bad

Unstated suggested conclusions:

Either – Yes, some NGOs have
sinister motives, to impose an
alien philosophy;
Or – Yes, they are just uncoop-
erative, ‘dissidents’;
So:
Either - We should impose
greater controls on them;
Or - to bring them into line with
us we might use allegations of
sinister alien plans

This text has less emphasis on a logical argument where one step builds from
previous ones, and more on conveying unstated conclusions by a series of rhetorical
questions for which the conclusions are implied by the choices of favour-
able/unfavourable words. Attention to the meanings of words used is therefore centrally
important. Scriven’s format remains very relevant for the questions it generates about
unstated assumptions and conclusions.

17
 Figure 8 transfers Figure 7’s insights about conclusions and assumptions, across
to a George columns diagram. There are three conclusions and so we need to map three
structures. The first row of the table indicates for each column a corresponding Scriven
step.

Figure 8: The logical structure of Minister Chikowore’s comments on NGOs

(Scriven step 2: find (part of step 3: find ba- (Step 4: identify as- (Step 6: consider other
conclusions) sic structure) sumptions) arguments)
I PROPOSE THAT GIVEN THAT AND THE UNLESS [REBUTTAL]
[CLAIM], [DATA] PRINCIPLE THAT
[WARRANT],

[1] Some NGOs have They act like missiles NGOs should be subor- The NGOs are guided by
sinister motives, to im- unguided by our people dinate [in the disputed our people in other ways;
pose an alien philoso- through their party and spheres] to the national and.
phy, or are simply unco- government and local governments ‘guidance’ is not the
operative [Unstated assumption] same as control.
[Conceivable counter-
[Unstated, but strongly arguments, against the
hinted-at, conclusion] claim or the supposed
data.]
[2] I am pro-NGOs I want the work of I want to dominate NGOs
[Unstated conclusion] NGOs to increase and that goes against
their basic rationale and
justification
[Conceivable counter-
argument to the claim]]
[3] Government would I am well-intentioned Government has the The ruling party and
be right to impose towards NGOs (conclu- right to control NGOs government are not the
greater controls upon sion [2] above) [Unstated assumption] sole repository of legiti-
NGOs, or to intimidate macy (e.g. since our peo-
them by allegations of And yet ple are not homogene-
sinister motives and/or ous).
threats to impose con- Some have sinister mo- Attempts to guide all
trols tives or are uncoopera- NGO action by others are
[Unstated conclusion] tive (conclusion [1] counter-productive.
above) [Conceivable counter-
arguments, against the
warrant or directly
against the claim]

Why are so many conclusions and assumptions unstated? Vagueness and veiled
communication are important in political language. The Minister can insinuate and hint
at threats, without committing himself irrevocably to either a definite action or even a
diagnosis. He can always later add counterarguments to the implied answers to his own
rhetorical questions, if and when that serves his purposes. He retains room for manoeu-
vre.

18
6. MIXED AND QUALIFIED MESSAGES: HARRY BLAIR ON RURAL
LOCAL GOVERNMENT IN INDIA
Harry Blair has a longer, multi-stage, more ambiguous story to tell.
"At first these elected local councils [in India after independence] were...taken
over by the traditional local elites, who used them to turn rural development pro-
grammes into patronage for themselves. But as time went on...the more numer-
ous middle-level peasants began to...win control [via] the ballot-box...[and to]
demand a larger rural development commitment from government... This whole
process in turn has led to a further awakening, as sharecroppers and landless ag-
ricultural workers have begun to demand government help... gentry and kulaks
struggle between themselves for power while trying simultaneously to co-opt and
beat back the...challenge from below" (Blair 1985: 455-6).

Blair gives suggestive descriptions of three stages in evolution and then of the
resulting situation. Figure 9 tries to clarify his interpretations, by looking closely at his
choices of terms in these four sentences. We use the same format as when investigating
Minister Chikowore on NGOs: a column for dividing the text into sections, a column
for examining the choices of terms and their implications, and a column for alternative
possible formulations of the same data, in order to highlight the judgements made by
this text and some of the counter arguments it will have to deal with.

19
 Figure 9: Step-wise analysis and paraphrase of Blair’s description of historical
stages
THE TEXT, divided into sec- COMMENTS, THE TEXT IN ALTERNATIVE
tions (steps 0, 1: components, On language (step 1) and implied LANGUAGE (step 1, step 6)
meanings; & part of 3: structure) conclusions (step 2)
[1] “At first these elected local ‘elected’ and ‘local councils’ are At first these local committees
councils [in India after independ- both relatively favourable terms; elected by the largely illiterate
ence] were...taken over by the ‘traditional local elites’ and electorate relied for leadership
traditional local elites ‘taken over by’ are not, therefore on the experience of leading lo-
there is an implied criticism of cal families who enjoyed local
phase 1 people’s trust and so were able to
win elections.

who used them to turn rural de- ‘rural development programmes’ These leaders were frequently
velopment programmes into pa- is a relatively favourable term, amongst those best placed to
tronage for themselves. and ‘patronage’ is generally benefit from the rural develop-
negative, so the implied criticism ment programmes (and naturally
of phase 1 becomes stronger, also gave priority to their close
perhaps even explicit supporters)

[2] But as time went on...the more ‘middle-level peasants’ sounds Decentralization worked even
numerous middle-level peasants good compared to ‘traditional better in its second phase, after
began to...win control [via] the local elites’; similarly, winning understanding of local govern-
ballot-box...[and to] demand a via the ballot-box carries kudos, ment had grown, and leaders
larger rural development com- and so does ‘larger rural devel- from larger population groups,
mitment from government ... opment commitment‘; so there is oriented to farming, began to win
implied praise of phase 2 elections.

[3] This whole process in turn has ‘awakening’ (which is a meta- In time, the democratic spirit has
led to a further awakening, as phor) sounds favourable; thus spread and appeared even in the
sharecroppers and landless agri- there is some implied praise of poorest groups, who have be-
cultural workers have begun to phase 3 come politically mobilized, vocal
demand government help... and active.

[4] ..gentry and kulaks struggle ‘gentry’ (French gentil = gentle) The traditional local leading
between themselves for power may have negative connotations families and the upcoming
while trying simultaneously to now; ‘kulak’ (Russian for fist, farmer groups inevitably com-
co-opt and beat back tight-fisted person; those peas- pete, within the democratic sys-
the...challenge from below" (Blair ants in pre-1930 Russia who tem. Both groups try to win sup-
1985:455-6). could generate substantial sur- port in the elections and to hold
plus) has clear anti- connota- on to key positions.
tions; and ‘co-opt’ and ‘struggle
for power’ are slightly negative;
whereas ‘challenge from below’
sounds progressive. Hence as-
pects of the current phase are
viewed with distaste and the im-
plied praise of phase 3 is
strengthened.

20
 Figure 10 now brings together the elements in Blair’s text, manifest or hinted at,
to see how they function as arguments, as well as some possible counter-arguments.
Why are the conclusions hinted at, rather than explicit? Commenting on massive, multi-
faceted, multi-potential, historical trends and contestations across a sub-continent, Blair
may not want to commit himself to highly emphatic claims.

Figure 10: Analysis of the structures of Blair’s suggestions

(Scriven’s Step 2: find Step (1 and) 3: find Step (1, 3,) 4: find un- Step 6: consider alter-
conclusions; support offered; stated assumptions; native arguments)
I PROPOSE THAT GIVEN THAT AND THE PRINCIPLE UNLESS
[CLAIM], [DATA] THAT [WARRANT], [REBUTTAL]

Phase 1 of elected local It was dominated by 1. Equity implies Traditional local elites
government in India was traditional local elites disadvantaged groups enjoyed full local ac-
unfortunate [Unstated who unfairly monopo- should be able to come ceptance
conclusion] lized programme re- forward. [Conceivable counter-
sources 2. Democracy implies that argument]
political office should not
be held for reasons other
than [x, y, z]; and should
not be used for one’s own
narrow advantage
[Unstated assumptions]
Phase 2 was an im- Larger groups mobilized [See 1. above] [Unless these middle-
provement, by compari- to win power by elec- peasant groups at the
son [Unstated conclu- toral means and cam- same time used political
sion] paigned for programmes power to increasingly
of relatively wide benefit exploit bottom groups]
[Conceivable counter-
argument]

Phase 3 is a further im- The poorest groups have [See 1. above]

provement become aware and mo-
[Unstated conclusion] bilized

Overall rural local gov- The trend noted earlier, The underlying logic that The poorer groups are
ernment is moving in the from phase 1 to phase 2 the access to political ignorant and vulnerable
right direction, even if to phase 3 office and the benefits and hence can be ma-
the process is not fast or available through win- nipulated and bought by
smooth or pleasant ning elections lead to the nasty gentry and
[Unstated conclusion] more power for the nu- kulaks
merically important [Counter argument that
poorer and disadvan- is hinted at]
taged groups
[Unstated assumption]

21
7. A MORE EXTENDED ARGUMENT: ‘THE STAR’ ON POPULATION
GROWTH IN SOUTH AFRICA
Can the method presented above for close analysis of arguments cope with
longer texts? Let us look at an editorial from ‘The Star’ of Johannesburg from the mid
1990s. It is almost four times as long as the extracts we analysed earlier, which already
proved quite intricate. We need to identify its main conclusion(s), and then place in per-
spective any sub-arguments, minor conclusions and digressions.

The China Syndrome

South Africa’s economy is being crippled by overpopulation
Our population is, almost Chinese style, increasing by 1-million a year –
a rate of growth that is entirely unsustainable and which is already defeating
measures to uplift South Africa’s black masses.
It is doubtful that any country, given our size and resources, could cope
in such circumstances. Ironically, it is often the very people who have no jobs
and who daily demand State assistance that are producing the most children –
children whom they can neither feed adequately nor clothe.
The reasons for the explosive population growth are well known: for a
start, the only assurance many rural people have of an income in old age, is
children – children who, they hope, will grow up to look after them. There are
other factors but one of the most serious is the mindless chauvinistic attitude of
unsophisticated males who father children, not because they care about children,
but simply to prove their own virility. The mothers and the children are too of-
ten abandoned to live off the State.
We have reached a stage where there are fewer and fewer workers hav-
ing to keep more and more non-workers. It is a recipe for disaster and one that
could reduce us to the level of just another African “beggar state” dependent on
foreign charity.
The Government is about to issue a position paper regarding population
growth. It will then seek public input and follow up with a conference to find

22
 solutions. One hopes the Green Paper, and the conference, will not balk at frank
discussion.
There are divisive cultural prejudices involved but unless we curb our
birthrate South Africa will not be able to survive within its environmental in-
come. We should note though, how the people of Italy and Ireland – countries
with ostensibly deeply rooted official prejudice against birth control – are in fact
attaining zero population growth.

As usual, one can start by trying to identify the different bits in the argument.
One needs to use a fairly fine-toothed comb in this first step. Here is my attempt. Ele-
ment [20] was separated out later than the others, when I started to order the elements
into asserted causes and effects.

[1] The China Syndrome

South Africa’s economy is being crippled by overpopulation
[2] Our population is, almost Chinese style, increasing by 1-million a
year – [3] a rate of growth that is entirely unsustainable [4] and which is already
defeating measures to uplift South Africa’s black masses.
[5] It is doubtful that any country, given our size and resources, could
cope in such circumstances. [6] Ironically, it is often the very people who have
no jobs and [20] who daily demand State assistance [6 cont.] that are producing
the most children – [7] children whom they can neither feed adequately nor
clothe.
[8] The reasons for the explosive population growth are well known: [9]
for a start, the only assurance many rural people have of an income in old age, is
children – children who, they hope, will grow up to look after them. [10] There
are other factors but one of the most serious is the mindless chauvinistic attitude
of unsophisticated males who father children, not because they care about chil-
dren, but simply to prove their own virility. [11] The mothers and the children
are too often abandoned to live off the State.

23
 [12] We have reached a stage where there are fewer and fewer workers
having to keep more and more non-workers. [13] It is a recipe for disaster and
one that could reduce us to the level of just another African “beggar state” de-
pendent on foreign charity.
[14] The Government is about to issue a position paper regarding popu-
lation growth. It will then seek public input and follow up with a conference to
find solutions. [15] One hopes the Green Paper, and the conference, will not
balk at frank discussion.
[16] There are divisive cultural prejudices involved but [17] unless we
curb our birthrate South Africa will not be able to survive within its environ-
mental income. [18] We should note though, how the people of Italy and Ireland
– [19] countries with ostensibly deeply rooted official prejudice against birth
control – [18 cont.] are in fact attaining zero population growth.

Second, one can examine the language in each part -- rich with value-laden
terms, often tendentious -- but more selectively in this case, given the much longer text.
That exercise is left for the reader. Third, we can draw a diagram or figure which shows
relations between the different parts. One may have to draw this two or three times to
reach what seems a coherent and balanced picture of the argument. The posited cause-
effect chains in some cases link six elements, but in Figure 11 I have summarized them
in four columns. Italicized elements are those unstated but implied in the editorial.
While a table lacks the exact indication of linkages given in a diagram by arrows, the
linkages are made sufficiently clear by the location of elements, and a table is far easier
to draw.
I judge that the main conclusion is implied by the editorial’s italicized subtitle:
‘South Africa’s economy is being crippled by over-population’. The editorial clearly not
only laments this but urges a policy response. Certain policy steps seem strongly im-
plied by the editorial’s listing of proposed causes and its strong, emotive, language --
‘crippled’, ‘mindless’, ‘divisive cultural prejudices’ [belonging to others, never to Star
editorial writers], and so on. But it is far less explicit here; it adopts, like Minister
Chikowore on NGOs, a politic cautiousness on specific measures.

24
 Figure 11: An informal structure diagram of the ‘China Syndrome’ editorial
POSITED CAUSES POSITED DATA POSITED POSITED POSITED POLICY
[8: ‘well-known’] EFFECTS EFFECTS IMPLICATIONS
(Earlier stages) (Later stages )
[9] ‘the only assurance [17] Environmen-
many rural people [21] Impacts on tally unsustainable
have of an income in environment- af- effects
old age, is children’ fecting activities
[10] Male chauvinism [2] Population [3] Generally un-
to prove virility growth of 1- [12] Rising de- sustainable effects
[16] (Other) ‘divisive million per year pendency ratio [1] South Africa’s [24] Government
cultural prejudices’ [22] Higher tax economy is being must act [Unstated
[11] Mothers and rates and less in- crippled by over- conclusion]
children (too) of- vestment] population]
ten abandoned [13] SA can be-
[6] Jobless often [20] Demands for come a beggar
produce most State assistance state
children; & [7] [4] Measures for
cannot adequately black uplift are
care for them undermined
[15] Government
[14] Govt. is about must be frank and
to issue a position bold
paper, etc.
[18] Italy & Ire- [23] Something can
land are attaining be done despite cul-
zero popn. growth tural prejudices
[19] despite offi-
cial prejudices
against birth con-
trol

Fourthly, one can clarify and focus this China Syndrome text by use of an RV
George table, as in Figure 12. Two stages are required: to show the cause-effect chain
(in fact one could use several stages for this if one wanted to trace the proposed cause-
effect analysis in more detail); and to show the implied subsidiary policy claim about
the desirability and feasibility of action.

25
 Figure 12: The argumentative structure of ‘The China Syndrome’’s main claims
I PROPOSE THAT GIVEN THAT [DATA] AND SINCE THE UNLESS
[CLAIM], RULE/PRINCIPLE [REBUTTAL]
THAT [WARRANT],
i) SA economy will be SA population is grow- The various posited in- Government takes a bold
crippled [and other ing at 1 million p.a. [and termediate effects/ link- lead with a new policy
claimed final effects] the other Data] ages connecting data to [or - counterargument -
claim unless the problem will
solve itself; examples of
Italy & Ireland]
ii) Government should Claim from part i., The posited underlying [The proposed action
act on the underlying causes [and intermediate would fail tests of feasi-
causes [and/or the inter- and the encouraging linkages] bility, efficacy, accept-
mediate links] examples of Italy and ability, etc., and/or of
Ireland comparison with alter-
natives, including the
[and tacit normative do-nothing alternative]
warrants]

The Toulmin format’s prompt to examine possible counter-arguments is helpful.

In this case it leads us to examine the brandished examples of Italy and Ireland, and to
ask: but did government act in those cases? Are they not examples of demographic
transition occurring in the absence of government policy?

8. CONCLUSION
The three main tools I have presented -- Scriven’s seven step model for investi-
gating an argument; the Toulmin model for describing structure, in the format provided
by George; and the study of praise/criticism language -- are relevant for examining ar-
guments from many types of field. Each however has some policy flavour or affinity:
Scriven is a leading theorist of program evaluation; the Toulmin model as adopted by
Dunn and other policy studies writers has been relatively influential; and the bestowal
of praise and criticism are central in policy language. A study module on policy argu-
mentation, and on policy discourse more broadly, should indeed though cover far more
than these three tools.
As mentioned at the outset, methods for analyzing complex systems of argumen-
tation (and other discourse) are out of reach for virtually all policy analysis practitioners
or students -- even for example the outstanding survey by van Eemeren et al. (1996) and
its Amsterdam siblings, the journal ‘Discourse and Society' and van Dijk (1997). Within
reach, and essential, are methods for analyzing and assessing policy arguments in par-

26
ticular: both more elementary methods, such as Hambrick’s schema, the logical frame-
work, and Dunn’s checklist for a policy issue paper (1981: 364-7); and some methods for
more complex policy discourses and 'policy frames', such as the work of Frank Fischer
(1995) and Emery Roe (1994, 1999). Roe’s work on ‘policy narratives’ and counter-
narratives is accessible and helpful: on the dangers of negativity in problem-oriented
policy analysis and the strong forces behind simplified policy stories (such as The
Star’s Malthusianism), and on suggestions for how to do better despite those forces, by
creating better even if still simplified stories (see e.g. Pain 1996). There are certainly
further relevant areas too (see e.g. other papers in Apthorpe & Gasper, eds. 1996). I
hope here to have whetted readers’ appetites, and to have at least provided some work-
able introductory tools to help students in their investigation of structures and mean-
ings.

REFERENCES
Apthorpe, R., & Gasper, D. 1982. Policy Evaluation and Meta-evaluation: The Case of
Rural Cooperatives. World Development, 10(8), 651-669.
Apthorpe, R., & Gasper, D. (eds.) 1996. Arguing Development Policy: Frames and
Discourses. London: Frank Cass.
Bozeman, B., and Landsbergen, D. 1989. Truth and Credibility in Sincere Policy
Analysis. Evaluation Review,13, 355-79.
Dijk, van T. (ed.), 1997. Discourse Studies: a Multidisciplinary Introduction. London:
Sage.
Dunn, W.N. 1981. Public Policy Analysis: An Introduction. Englewood Cliffs, NJ:
Prentice Hall.
Dunn, W.N. 1994. Public Policy Analysis: An Introduction. 2nd edition, Englewood
Cliffs, NJ: Prentice Hall.
Eemeren, F. van, et al. 1996. Fundamentals of Argumentation Theory. Mahwah, NJ: Law-
rence Erlbaum Associates.
Fischer, F. 1995. Evaluating Public Policy. Chicago: Nelson-Hall.
Gasper, D. 1991: Decentralization of Planning and Administration in Zimbabwe -
International Perspectives and 1980s Experiences; pp.7-50, The Limits of

27
 Decentralization in Zimbabwe, A. Helmsing et al., The Hague: Institute of Social
Studies.
Gasper, D. 1996a. Analysing Policy Arguments. In Apthorpe & Gasper (eds.), pp. 36-
62.
Gasper, D., 1996b. Essentialism In and About Development Discourse. In Apthorpe &
Gasper (eds.), pp. 149-176.
Gasper, D. 1997. ‘Decentralization', ‘Coordination', ‘Integration' - Reflections from
Zimbabwe in the 1980s. Pp.51-71 in Locality, State and Development - essays in
honour of Jos G.M. Hilhorst, edited by A. Helmsing & J. Guimaraes, The Hague: In-
stitute of Social Studies.
Gasper, D. 1997. The Logical Framework Approach: a Critical Assessment. Working
Paper 264, The Hague: Institute of Social Studies.
Gasper, D., 2000. Evaluating the “logical framework approach” - towards learning-
oriented development evaluation, Public Administration and Development, 20( ).
Gasper, D., & George, R.V. 1998. Analyzing Argumentation in Planning and Public
Policy: Assessing, Improving and Transcending the Toulmin Model. Environment
and Planning B: Planning and Design, vol. 25, pp.367-390.
Goldstein, H.A. 1984. Planning as Argumentation. Environment and Planning B:
Planning and Design, 11, 297-312.
Pain, A. 1996. Rereading ‘Mountainous’, ‘Isolated’, ‘Inaccessible’ and ‘Small’: The
Case of Bhutan. European Journal of Development Research 1996(1), 63-76
Ray, E. 1990. Policy Evaluation and the Argumentation Process. Pp.61-73 in S.S. Nagel
(ed.), Policy Theory and Policy Evaluation, Westport, CT: Greenwood Press.
Roe, E. 1994. Narrative Policy Analysis. Durham, NC: Duke Univ. Press.
Roe, E. 1999. Except-Africa, New Brunswick, NJ: Transaction Publishers.
Scriven, M. 1976. Reasoning, New York: McGraw-Hill.
Sillince, J. 1986. A Theory of Planning, Aldershot: Gower.
Toulmin, S. 1958. The Uses of Argument, Cambridge: Cambridge Univ. Press.
----, et al. 1979. An Introduction to Reasoning, New York: Macmillan.

28
1
In English, ‘argument' can be 1. a single move from premises to conclusion, or 2. a whole series of
moves, or 3. a dispute. ‘Argumentation' is a) the activity of arguing, making arguments, or b) a linked set
of single arguments. The latter can also be called a ‘discourse'. ‘Discourse' in linguistics means any
stretch of language longer than a sentence. Not all stretches have an argument, e.g. they may be purely
descriptive. On other uses of the terms ‘discourse' and ‘discourse analysis', see Gasper & Apthorpe
(1996).
2
These last two roles are best indicated through illustrations, for example an exercise in structuring a new
argument such as this one: “Using all six elements described in the [Toulmin-Dunn model], diagram a policy
argument that results in a normative claim. Begin your argument with the following policy-relevant informa-
tion: ‘In larger municipalities (over 25,000 persons), the greater the number of families below the poverty
line, the greater the number of reported criminal offences’. Be sure to include a rebuttal in your argument
and pay careful attention to the way you formulate the problem.” (Dunn 1994:87).
3
The study drawn on for this section and the next, Gasper (1991 & 1997), provides illustrations from
decentralization discourse of: (a) examining meanings by a dimensions analysis, rather than just listing
variants, since many important concepts are vectors not scalars; and (b) examining metaphors and images,
and how they can mislead (e.g., ‘decentralization’ is not just the opposite of centralization, and ‘bottom-
up’ is not just the inverse of ‘top-down’). In his attack on NGOs (Section 5 here) Minister Chikowore
tends towards corporatist language: ‘our people through their party’; a society is discussed using the un-
derlying image of a body; to be ruled of course by its one head.

29