Papers

Ian Denley & John Long

Formerly: Ergonomics and HCI Unit, University College London, 26 Bedford Way, London WC1H OAP, UK.

e-mail: ian.denley@systemc.com; j.long@ucl.ac.uk

Abstract

This paper develops an approach that supports multidisciplinary practice in requirements engineering.

It is argued that multidisciplinary requirements engineering practice is ineffective, and some specific problems for multidisciplinary practice are identified. It is also suggested that the incommensurability of conflicting paradigms is an underlying cause of the problems in multidisciplinary practice, and a number of criteria for support to overcome such problems are proposed.

A form of methodological support, which it is claimed may help overcome some of the problems associated with multidisciplinary practice in requirements engineering, is developed. This methodological support takes the form of a dialectic process, and its associated products, which is conceptualised and then operationalised.

As an illustration of the methodological support offered to multidisciplinary practice, the operationalisation of the dialectic process is applied to requirements constructed by the use of two different requirements engineering techniques from two different disciplines (representing two different paradigms), in the domain of Accident and Emergency healthcare.

1.  Introduction

This paper develops an approach that supports multidisciplinary practice in requirements engineering.

In Section 2, it is argued that multidisciplinary requirements engineering practice is ineffective, and some specific problems for multidisciplinary practice are identified. It is also suggested that the incommensurability of conflicting paradigms is an underlying cause of the problems in multidisciplinary practice, and a number of criteria for support to overcome such problems are proposed.

In Section 3, a form of methodological support, which it is claimed may help overcome some of the problems associated with multidisciplinary practice in requirements engineering, is developed. This methodological support takes the form of a dialectic process, and its associated products, which is conceptualised and then operationalised.

Section 4 illustrates informally how the dialectic approach might meet the criteria proposed in Section 2.

Section 5 provides an illustration of the methodological support offered to multidisciplinary practice, the operationalisation of the dialectic process is applied to requirements constructed by the use of two different requirements engineering techniques from two different disciplines (representing two different paradigms), in the domain of Accident and Emergency healthcare.

2.  Multidisciplinary Practice in Requirements Engineering: Problems and Criteria for Support

This section is concerned with multidisciplinary practice (MDP) in the requirements engineering (RE).  It argues that multidisciplinary requirements engineering practice is ineffective. MDP in RE is introduced, exemplified, and its problems identified the incommensurability of conflicting paradigms may be a possible underlying cause of these problems.  The accommodation of alternative paradigms is discussed, and criteria for support to overcome the problems are proposed.

2.1  Multidisciplinary RE

RE remains an emerging field, and there is little consensus as to its nature and scope.  The emergence is characterised by a proliferation of perspectives, and a profusion of concepts, and it has been suggested for some time that RE may be multidisciplinary in nature and includes contributions from a wide range of disciplines including software engineering, human factors, cognitive psychology, and sociology (Scaife et al., 1994).

Surprisingly little attention has been paid to the nature of MDP per se.  The next section has the following aims: to distinguish between possible types of MP in RE; to illustrate and amplify them; and to review some examples.

2.2  Types of practice involving multiple disciplines

We distinguish between two general types of practice:

• Independent MDP – each contributing discipline addresses its own RE problem without influence on or by, other disciplines.
• Dependent MDP – contributing disciplines exert a mutual influence upon one another

2.2.1  Independent MDP

In independent MDP, the contribution provided by one discipline and its practitioner does not affect the contribution of other disciplines and their practitioners.  Each discipline addresses its own RE (See Figure 1).

Figure 1: Independent Multidisciplinary Practice in RE

Supposing the directors of a hospital are re-designing their Accident and Emergency (A&E) department to cope with increasing demands and diminishing resources.  This redesign might consist of:-

• Redesign of the department’s organisation and management structures
• Procurement of a new information system.

If the re-design is carried out by independent MDP, the organisational structures could be re-designed without consideration of the new information system, and vice versa.  Practitioners of the discipline of organisational/management psychology (management consultants) might identify a requirement to change the hierarchical structure of A&E staff to multi-professional team-working.  However, practitioners of the discipline of software engineering (SE – specialist software developers) might identify a requirement to design the A&E information system around an official NHS minimum data set for A&E (including generic tasks such as registering patients; triaging patients etc.) without consideration of a new team-working structure.

Knowledge and practices from the discipline of organisational/ management psychology would be applied independently of knowledge and practices from of SE.

2.2.2  Dependent MDP

In dependent MDP, the contribution to the RE process provided by one discipline and its practitioners occurs in the context of the other disciplines and their practitioners also offering contributions.  Disciplines exert a mutual influence upon one another (see Figure 2).

Figure 2: Dependent Multidisciplinary Practice in RE

In system development, the disciplines of human factors (HF) and SE exert a mutual influence on the design of behaviours at the user interface.  Similarly, the disciplines of HF and sociology exert a mutual influence on the design of behaviours for multi-user systems.

Extending the A&E example, imagine the hospital directors have identified a business requirement to introduce electronic communication between the A&E department and the pathology laboratory, because paper-based investigation requests often go ‘missing’.  Using knowledge and practices (task analysis), the system developer identifies that doctors currently fill-out and sign the paper-based investigation request forms, including all data.  The developer suggests this task should continue to be allocated to the doctor, and suggests that doctors also complete the electronic investigation request forms, including all data.  However, an ethnographic study by the system developer discovers that A&E doctors perceive data input tasks to be administrative (and to detract from caring for patients), and only use the computer for information retrieval. The doctors consider data input should be performed by nurses. The system developer cannot simply utilise the two discipline contributions independently of one another.  Instead, the developer must understand how the contributions are related, and how they should influence each other in a requirement for the design of the investigation ordering system.

2.2.2.1 Types of Dependent MDP

How might dependent MDP be carried out? We identify four types of dependent MDP: by concept; by product; by process; by practitioner.

MDP by concept occurs if the very foundations of disciplines 1 and 2 were integrated within a common framework or theory (a hybrid discipline).  No such theories or frameworks exist within RE at present.

MDP by product occurs when a practitioner of discipline commissions a requirements study from a practitioner of Discipline 2, and receives a product (the results of an analysis) to be considered with products of their own discipline. MDP is organised around the use of substantive discipline knowledge.  Sommerville et al. (1998) use a viewpoints framework to make explicit the relationships between multiple discipline perspectives in the requirements for an on-board train protection system.

MDP by process occurs if a practitioner of Discipline 1 learns a technique from Discipline 2 and incorporates it into his/her repertoire.  MDP is organised primarily around the use of methodological knowledge.  This type of MDP is often advocated.  Macaulay (1996) suggests requirements engineers should acquire a range of RE techniques; and Sutcliffe (1997) also proposes a technique combination approach.

MDP by practitioner occurs if a practitioner of Discipline 1 collaborates with a practitioner of Discipline 2.  Both practitioners use their training and judgement to respond to specific concerns.  The practitioners themselves act as channels for their respective discipline knowledge.  This paper is primarily concerned with dependent MDP by practitioner.

2.3  Examples of Dependent MDP by Practitioner

Three case-studies of MDP in RE were examined, involving collaboration between the disciplines of: psychology and SE; sociology and SE (Denley, 1999).  Each case is an example of dependent MDP by practitioner.

2.3.1  Psychology and Software Engineering

Gasson (1995) studied MDP between psychologists and software engineers to develop computerised support interactive student learning.  The study analysed design documents from the development team (psychologists and software engineers), and interviews with team members.  The project plan required the team to follow an integrated development process model.  This model aims to support collaboration.  This RE model failed almost immediately.  Gasson speculates there was a dichotomy between the two disciplines, and team members from both failed to understand the requirements of the other or to learn from each other.

Scaife et al. (1994) report a multidisciplinary project aimed to develop a software tool for fashion designers.  The development team comprised software engineers and psychologists/ cognitive scientists. The Study focuses on some of the difficulties of software development arising from their interaction. Such difficulties were marked in the early stages of the project, during the translation of empirical data requirements into specifications.  The case study identified a number of specific problems including: a lack of shared assumptions about the development process; problems understanding terminologies across disciplines; and methodological differences about the ‘correct’ way of working.

2.3.2  Sociology and Software Engineering

This case-study concerns the requirements for a user interface to a flight database providing real-time confirmation to air traffic controllers.  The study was constructed from four reports of the project; two from the software engineers (Sommerville et al., 1993); and two from the sociologists’ perspective (Hughes et al., 1993).  This case-study also represents an example of MDP by practitioner. Multidisciplinary interactions were organised around monthly debriefing meetings.

Both the sociologists and the software engineers retain positive views on their collaboration.  However, both groups report problems.  There is a wide gulf between their disciplines due to entrenched philosophical positions.  The case study identified a number of specific problems, including; communication problems due to different terminologies; methodological differences; and problems understanding ‘what the other discipline practitioners actually do’.  Future successful collaboration would require both disciplines to question further their own assumptions and working methods.

2.4  Problems in Dependent MDP by Practitioner

The case studies examined above all identify similar problems with dependent multidisciplinary practice by practitioner. These include:

• dichotomy of approach between the two disciplines
• problems of viewpoint, and the different interests that the disciplines take in the world
• lack of shared assumptions (about the system development process)
• irreconcilable differences between alternative methodologies
• the failure of team members to understand each other’s disciplines or learn from each other
• problems in understanding what the other discipline practitioners actually do
• the need for each discipline to make undesirable compromises
• problems with understanding terminologies across disciplines
• entrenched philosophical positions

In sum, even though we have only examined a small sample of case-studies, there appears to be serious obstacles to effective MDP by practitioner.  Next we explore their possible causes.

2.5  Dependent MDP: The Problem with Paradigms

To understand such problems, we need first to understand disciplines.  The notion of a paradigm (Kuhn, 1962; 1970; Burrel & Morgan, 1979; Guba & Lincoln, 1989; 1994) is used to guide reasoning about the nature of disciplines in terms of the types or problems they address, the practices they employ and the knowledge that supports those practices (Long & Dowell, 1989).  Paradigms can be understood as fundamentally different sets of assumptions about the nature of the world and how to obtain knowledge about it.  Paradigms are the bedrock on which disciplines are built.

Furthermore, a possible underlying cause of problems in dependent MDP is that paradigms are philosophically incommensurable.  Paradigms have no common measures, and cannot be mixed together (Jackson & Carter, 1991).  However, dependent MDP by practitioner may only be possible if we concede the possibility of the practical accommodation of alternative paradigms.  Finally, some criteria are proposed which, if met, would support such practical accommodation.

2.5.1 Disciplines and Paradigms

Long & Dowell (1989) suggest that most definitions of disciplines assume three primary characteristics: knowledge; practices; and a general problem with a particular scope.  The scope of general (discipline) problems can also be decomposed resulting in sub-disciplines.

The division of a discipline into sub-disciplines is useful.  It allows us to classify disciplines and examine their commonalities and differences. The discipline of science is a sub-discipline of a super-ordinate discipline whose scope might include metaphysical beliefs about the nature of the world, the human individual’s place in it, and possible relationships between human beings and the world.  Such a discipline addressing sets of basic assumptions and beliefs is known as a paradigm (Burrel & Morgan, 1979; Guba & Lincoln, 1989; Guba & Lincoln, 1994).  In their perspective, a paradigm may be viewed as a set of basic beliefs that represent a world-view defining for its holder the nature of the ‘world’, the individual’s place, and relationships to that world.  The beliefs are basic, in that they can only be accepted on faith (however well the beliefs are argued).  There is no way to establish their ultimate truthfulness.  They define a paradigm as:

“the basic belief system or work-view that urges the investigator, not only in choices of method but in ontologically and epistemologically fundamental ways.”

These basic beliefs are summarised by how their proponents respond to three fundamental questions.  The answer to any one question, taken in any order, constrains answers to the others.

• Ontological question: what is the form and nature or reality, and therefore, what can knowledge be about?
• Epistemological question: what is the relationship between the knower and what can be known?
• Methodological questions: how can the knower find out about whatever he or she believes can be known?

Guba & Lincoln (1994) believe the order of these questions reflects a logical, if not necessary, primacy.  They also argue that they can be answered in different ways, and each formulation constitutes a different paradigm.  Answers to these questions cannot be proven. They further argue that answers given are human constructions; all inventions and hence subject to error.  Advocates of any particular construction must rely on its persuasiveness and utility rather than proof.

2.6  Discussion on the Accommodation of Paradigms

Despite their differences, all three earlier perspectives share a common feature: they all believe paradigms are philosophically incommensurable, that is, having no common measure, and cannot be mixed together (Jackson & Carter, 1991; Sankey, 1994).  Conversely, a re-occurring theme is the possibility of achieving some other kind of accommodation between them at a practical level.

2.6.1  Incommensurability of Paradigms

That paradigms are distinct and cannot be mixed is based on the notion of irreconcilable conflicts between their philosophical positions on certain key features including: criteria; facts; meaning; and values – after Kitchener (1986).  Each feature is considered.

Criteria: Competing paradigms select different problems, as the most important to solve, and employ different standards, against which to judge the success of the solution.  Incommensurability is typically taken to imply no common measures among paradigms of inquiry; rival paradigms cannot be comparatively evaluated by a neutral set of rules or criteria (Lincoln, 1990).

Facts: Rival paradigms do not share a common body of data providing a neutral standard for their comparison; their fundamentally different theoretical viewpoints lead to different perceptions and interpretation of the facts.  There are no neutral facts, free of all theory and empirically given.  All facts are relative with respect to particular paradigm. Similarly, the content of paradigms cannot be compared, since they are logically and epistemologically incompatible.

Meaning: Paradigms cannot be compared because there is no language independent of theory.  Languages categorise the world in different ways, and in the transition between paradigms words change their meanings or conditions of applicability (Kuhn, 1970).

Values: Guba & Lincoln (1994) adopt an unashamedly relativistic position, when they say that all paradigms are human constructions products of the human mind and subject to error.  For them:

“the basic beliefs of paradigms are essentially contradictory . . . value freedom and value boundedness, cannot coexist in any internally consistent metaphysical system.”

In summary, then, paradigms are incommensurable since the rules for action, for process, for discourse, for what is considered knowledge and truth, are different and lead to diverse, disparate, distinctive and typically antithetical ends (Lincoln, 1990).

Given the incommensurability of paradigms, some authors suggest that it may be possible to accommodate different paradigms within multi-disciplinary practice within a given field – see, for example, (Weaver & Gioia, 1995) for organisational studies and (Chua, 1986) for accountancy. Such practical accommodation might take place either through the collaborative work of groups of practitioners, (social-community accommodation), holding different paradigms, or via the practice of individuals (individual accommodation) – after Skrtic (1990).

2.6.1.1  Social-community Accommodation of Paradigms

The implication outlined earlier is that a scientific community is premised on meta-theoretical assumptions, defined by a paradigm. Without a paradigm, there can be no scientific community (Skrtic, 1990).  We still might ask whether advocates of alternative paradigms can live with (and learn from) each other?  – after Austin (1990).

The emphasis, then, of the social-community accommodation is on communication between different schools of thought.  Accommodation involves understanding the key issues of each paradigm, appreciating the different views of others, allowing multiple contributions to the same inquiry (constructing requirements for the redesign of an information system, for example).

However, this type of accommodation has its dangers, both for practitioners and paradigms.  First, social-community accommodation may be more apparent than real. Multi-disciplinary teams are ways to provide differed perspectives.  Their differences produce a wide range and diversity of points of view. However, apparently different theorists may be located within the same paradigm even if they do not share identical perspectives (Burrel & Morgan, 1979).  Socio-technical theory still operates within a traditional functionalist paradigm, despite its social perspective.  Multi-disciplinary teams, therefore, often fail to given an all-round view.  The second danger is incorporation.  Alternative paradigms may be regarded as simply points of view to be considered, and if possible, rebuffed or incorporated within the dominant orthodoxy.  Such a view favours incorporation as the line of intellectual development.

The third danger is that non-dominant paradigms are forced to take up the role of handmaiden to the dominant paradigm (Guba & Lincoln, 1989).  This situation occurs when an alternative paradigm is used to make good the deficiencies of the dominant one without calling that one into question.  An ethnomethodological approach to inquiry may be taken onto a conventional software development process to unearth social issues of interest that can be passed onto the software engineer to inform the ‘real design work’.

Despite these dangers, the social-community accommodation of paradigms has been pursued in RE.  Macaulay (1996) proposes to improve the communication process between different disciplines in the early phases of information systems design.  A number of requirements methods attempt to integrated methods from different paradigms, e.g. MEASUR (Stamper, 1994).  All these methods attempt to integrate a social/ organisational perspective (such as  Checkland’s (1981) soft systems methodology) with a software engineering perspective (such as JSD or Object Oriented Analysis).

Finally, although for social-community accommodation it is not necessary for each researcher to be able to operate within different paradigms, each researcher should appreciate other views from alternative paradigms.  However, the individuals involved may not even be able to reach this level of appreciation.

The depth of the paradigm socialisation processes leads to a fourth danger.  Each paradigm may be sufficiently divergent, and the emotional and political commitments so high, that accommodation within either research or practice, may produce only dissonance and incoherence (Lincoln, 1990).

In sum, then, we conclude the social-community accommodation of paradigms has a number of dangers that need to be appreciated, and avoided, for communication between paradigms to be successful.

2.6.1.2  Individual Accommodation of Paradigms

A further implication from earlier arguments is that the socialisation of individuals into accepting the assumptions of a paradigm from another. Paradigms and their communities may constrain the behaviour and beliefs of their individual members.  Such constraints may lead individuals, having only limited exposure to alternative perspectives, to conclude that theirs is the only way to think and operate.  Representatives of different paradigms, then, may live in different worlds, hold mutually exclusive beliefs, use different vocabularies and have different ways of seeing.  Despite such differences, however, we might still ask whether, “an individual researcher or practitioner can accommodate various paradigms?” – after (Austin, 1990).

Firestone (1990) believes that paradigm accommodation depends on one’s stance on the nature of paradigms, the philosophical principles and research practice.  Those who argue that paradigms are incompatible view them as systems of rules that are largely deductive; where assumptions about the nature of the world (ontology), and how one knows it (epistemology), govern the conduct of the research.  Firestone suggests an alternative conception that views the paradigm-practice relationship as bi-directional, rather than uni-directional.  In practice, researchers use a variety of imperfect approaches to enhance the credibility of their arguments.

In short, Firestone believes that an accommodation of paradigms, is possible in the level of the individual, and argues that individuals should let the various paradigms inform practice to the extent that they are useful.  The methods that characterise paradigms can be combined by the individual to solve particular problems and answer particular questions.  With respect to RE, Macaulay (1996) suggests that a requirements engineer might develop a portfolio of techniques which can be drawn upon, depending on the given situation, and be used in combination as required.

Although Firestone’s bi-directional understanding opens up the possibility of personal accommodation, few practitioners consciously articulate the paradigms from which they borrow techniques.  They are unlikely to understand the implications of using such techniques.  Such practitioners are content to live with internal inconsistencies that they neither recognise nor even particularly value (Crandall, 1990).

Individual accommodation of paradigms, then, requires that the role of the paradigm be brought into the realm of conscious thought.  Schon (1983) suggests that such a conscious use of paradigms may be a difficult, but not an impossible task.  Given a sufficient understanding of alternative paradigms, then, an individual may behave congruently with selected paradigms.

Lincoln (1990) believes that learning to use and possibly adapting a new paradigm is an intensely personal process, involving not only from intellectual, but also personal, social and possibly political transformations.  Reinharz (1981) has developed a model of the process by which individuals develop a commitment and ability to operate within a different paradigm.

We conclude that individuals may be able to operate in more than one paradigm, but that such practical accommodation can only take place over time, either by conversion or critical reflection.

2.7  Criteria for Support for Dependent MDP in RE

Earlier, we explored the general notion of dependent MDP in RE, and have highlighted a number of problems with dependent MDP by practitioner (see Section 2.4).  We have argued that the root cause of such problems lies in the philosophical incommensurability of paradigms.

We propose that dependent MDP by practitioner relies upon the practical accommodation of alternative paradigms.  Overcoming the problems of dependent MDP by practitioner relies upon: communication between different ‘schools of thought’ (i.e. social community accommodation); and the need for individual practitioners consciously to articulate, reflect upon, and perhaps even operate within, alternative paradigms (i.e. individual accommodation).

In Table 1, we have condensed the discussions of this section into a set of non exhaustive criteria which, if met, would support the practical accommodation of paradigms and hence support dependent multidisciplinary requirements engineering practice.

A. Criteria for the social-community accommodation of paradigmsAny approach that aims to support the practical accommodation of paradigms should:1) Support multiparadigmatic contributions to the same inquiry (Austin, 1990).2) Help practitioners understand the key issues of each paradigm (Crandall, 1990) 3) Help practitioners appreciate the views of those operating in a different paradigm (Austin, 1990). 4) Allow each paradigm to express itself independently of the other (Burrel and Morgan, 1979). 5) Help improve communication between practitioners of different paradigms (Guba and Lincoln, 1989) 6) Help practitioners understand the practical implications of alternative paradigm positions (Skrtic, 1990). 7) Help practitioners understand alternative beliefs (Burrell and Morgan, 1979). 8) Help practitioners understand the values and norms of alternative paradigms (Schön, 1983). B. Criteria for the individual accommodation of paradigms Any approach that aims to support the practical accommodation of paradigms should: 9) Bring the role of the paradigm into the realm of conscious thought (Schön, 1983). 10) Support practitioners in the conscious articulation of alternative paradigm positions (Crandall, 1990) 11) Help practitioners change their own behaviour and beliefs (Le Compte, 1990). 12) Help practitioners use different vocabularies (Barbour, 1980; Jackson and Carter, 1991) 13) Support practitioners in the enumeration of specific paradigm positions and criticisms (Reinharz, 1981). 14) Support practitioners in the management of commitment and critical reflection (Reinharz, 1981).

Table 1: Criteria for an approach to support the practical accommodation of paradigms

In this section, we have suggested that overcoming the problems associated with dependent MDP by practitioner depends upon the practical accommodation of alternative paradigms.  We have proposed a number of criteria that any type of support (such as methodological support) should aim to meet in order to facilitate the practical accommodation of paradigms, and hence overcome some of the paradigmatic constraints acting on dependent MDP.

In the next section, we will develop a dialectic approach that it is claimed can support dependent multidisciplinary practice by practitioner in requirements engineering.

3.  Dialectic Approach to Supporting Dependent Multidisciplinary Practice by Practitioner

In the previous section, we suggested that multidisciplinary practice by practitioner depends upon the practical accommodation of paradigms. Two types of accommodation were identified: social-community accommodation and individual accommodation. We also proposed that a dialectic approach may have the potential to support these two types of accommodation. This chapter develops such a dialectic approach.

Specifically, in this section, we develop a dialectic process which we then operationalise using a framework for argumentation that supports reasoning about the influence of the outputs of one paradigm/discipline upon another (in the context of requirements engineering).

We then illustrate how the dialectic process allows us to reason about the mutual influence of two requirements engineering techniques (one from the discipline of human factors, and one from the discipline of sociology; representing the post-positivist and constructivist paradigms respectively).

The aim of this section is to provide a ‘first pass’ at developing a dialectic approach that can support multidisciplinary practice by practitioner by supporting the practical accommodation of paradigms.

Section 3.1 introduces briefly a number of concepts associated with dialectic reasoning, reviews previous literature in the area, and considers the benefits and limitations of a dialectic approach. Section 3.2 develops a specific conceptualisation of the dialectic process to be used in this paper, and Section 3.3 operationalises this conceptualisation, using a framework for describing and analysing arguments. Sections 4 and 5 show how we can use the dialectic process and the framework to support multidisciplinary practice by practitioner in the field of requirements engineering.

3.1  Brief Introduction to Dialectic Reasoning

This section introduces the basic concepts and ideas associated with dialectic reasoning.

Dialectic reasoning is an ancient form of thinking which deals with contradiction. In its most general sense, however, dialectic reasoning has come to signify a process which analyses conceptual, social (and even natural) conflicts and contradictions with the aim of transcending them (Bhaskar, 1993, 1994).

Dialectic reasoning has frequently been used in the field of organisational decision-making and economic planning. Much of the work in this field has been stimulated by Churchman’s exploration of philosophically-based information systems (summarised in Churchman, 1971).

Mason (1969), for example, suggested that a dialectic approach to decision making involved examining an organisational problem from two opposing points of view, using a structured debate which consists of a forceful presentation of two opposing plans. Mason termed this process dialectical inquiry (DI). He investigated the effects of DI on strategic economic planning in a manufacturing company, and concluded that the managers who used the technique formed a new more encompassing conceptualisation of the strategic planning problem (leading them to adopt a policy that they had not originally considered). Mason’s work was extended by Mitroff (1971) who developed a mathematical model of DI based on probability theory, and Mitroff et al (1979) who used DI with a pricing decision in a drugs company. Three groups of managers, each advocating a different pricing policy, used DI to examine their divergent assumptions and arrive at a final policy. The authors reported that DI produced more and better alternatives and led to a different pricing policy than would have been chosen if DI had not been used.

Mason and Mitroff (1981) subsequently combined their earlier efforts into a technique for bringing out the underlying assumptions of any policy decision. This technique, the strategic assumption surfacing technique, incorporated dialectic debate procedures and a theory of argumentation which included a mathematical formulation of Toulmin’s (1958) classic model of argumentation. This paper builds on the work on Mason and Mitroff in its use of an extended and enhanced formulation of the Toulmin model (see Section 3.3).

In sum, there seems to be reasonable evidence that dialectic reasoning (as represented in this instance by DI) supports the identification of, and reasoning about, the assumptions underlying opposing positions. There is also somewhat more limited evidence that it may also be useful in improving the quality of strategic decision-making. On this basis, it seems reasonable to suggest that dialectic reasoning might have a potential role to play in supporting the practical accommodation of alternative paradigms (at least insofar as such accommodation relies upon communication between different ‘schools of thought’ and the conscious articulation of, and reflection on, alternative paradigm positions – see Section 2.6). This suggestion is explored in more detail in the remainder of this chapter.

3.2 Specific Conceptualisation of the Dialectic Process and its Products

The studies mentioned in Section 3.1, do not employ a specific conceptualisation of the dialectic process and its associated products. This section develops such a conceptualisation, and then Section 3.3 operationalises this conceptualisation in the form of a framework within which to model (that is to describe and analyse) argumentative reasoning (of the sort required by the dialectic process). Section 4 argues that this conceptualisation and its operationalisation can provide support for the accommodation of alternative paradigms.

Much contemporary discussion of the dialectic approach revolves around the figure of Hegel, and his two treatises The Phenomenology of Spirit and The Science of Logic.

The Phenomenology might be crudely summarised as proposing that no philosophical viewpoint or conceptual framework, no matter how persuasive, can ever be adequate by itself. For every premise or set of premises, there is a context and a set of presuppositions that has been taken for granted; for every argument there is a perspective that has not been challenged; for every ethical argument there is a social, cultural and historical context that renders it intelligible.

In the Logic, the dialectic is understood as a method for expounding fundamental categories, including forms of judgement and argument as well as fundamental concepts (Forster, 1993). The dialectic method was a process which remedied one-sidedness and incompleteness in arguments, and resulted in the reconciliation of contradictions. The dialectic process progresses in two ways: first, by bringing out what is implicit, but not explicitly articulated in some notion; and second, by repairing some inadequacy in it.

Figure 3.0 provides a specific conceptualisation of the dialectic process and its associated products. In the figure, processes are represented by labelled arrows, and products are represented by boxes.

Figure 3 : Specific Conceptualisation of the Dialectic Process and its Products

The component parts of the dialectic process and its associated products are considered below.

Pre-reflective knowledge – Hegel conceived of the pre-reflective reasonableness of ordinary life which readily tolerates contradictions without finding anything problematic in them. This state of being can be transformed by a process of reflection which includes the development of self knowledge and discipline knowledge through learning, education and training, and discipline socialisation leading to a more informed state known as understanding.

Understanding is the state in which an individual can take part in the theory and practice which characterise his or her discipline. Understanding in this sense might be thought of as a position from which one can engage in normal science (in the Kuhnian sense). However, if we stay at the level of understanding, we will not find or recognise contradictions in our concepts (beliefs/theory) or experience (practice), and to move beyond this stage takes considerable effort via the process of negative reason.

Negative reasoning stretches our concepts to the limit by forcing contradictions out of them and pressing contradictions upon them (Bhaskar, 1994). Negative reasoning is characterised by the construction of the strongest possible debate or disagreement on a given issue (Mitroff, 1971). The end point of the process of negative reasoning results in a dialectic comment on the practice of the pre-existing community which reveals and exposes its inadequacies and inconsistencies. In Kuhnian terms, a dialectic comment might have parallels with the epoch of scientific revolution. In dialectic reasoning, the dialectic comment is the actual or notional moment which mediates between the negative reasoning and positive reasoning.

Positive reasoning can be characterised as creative speculation that leads to the expansion of the existing conceptual field. If reflection is the act of reflecting on our own subjective views or interests, speculation is the act of attempting to free oneself from these biases in order to work through the conflicting elements identified in negative reasoning (Hodgson, 1993). Expansion of the existing conceptual field (or world view) through positive reasoning results in the transformation of contradictions that leads ultimately to a resolution of the conflict (Cosier et al, 1978).

Resolution is characterised by the incorporation of contradictions into a fuller, richer, more comprehensive understanding; and is followed by a ‘return to life’. This return to life takes us to a new starting position from which the dialectic process can begin again.

In this paper, we are concerned primarily with the three processes of reflection, negative reasoning, and positive reasoning, and their respective products. The processes and products of concern here are shown in Table 2

Process	Product
Reflection	Understanding
Negative Reasoning	Dialectic Comment
Positive Reasoning	Resolution

Table 2: Dialectic Processes and Products of Concern

The next section considers how these dialectic processes and products may be operationalised.

3.3  Operationalising the Dialectic Process: Framework for Analysing Arguments

The dialectic process shown in Figure 3 and Table 2, is operationalised below using a framework which provides a conceptual structure within which to model (that is to describe and analyse) argumentative reasoning (of the sort required by the dialectic process). That is, the framework operationalises three key features of the dialectic process, namely: reflection, negative reasoning and positive reasoning (see Section 3.2) and their associated products. The framework comprises:

i) a general conceptual model of argumentation (after Toulmin, 1958; Toulmin, et al 1984)

ii) criteria for assessing the cogency of arguments

These two components of the framework are discussed in Sections 3.3.1 and 3.3.2 below.

In Section 4, it is claimed that the dialectic process and its operationalisation supports the practical accommodation of paradigms (and hence supports multidisciplinary practice by practitioner in the field of requirements engineering).

3.3.1  General Conceptual Model of Argumentation

The dialectic process described in Section 3.2 is predicated upon the need for reflection, negative reasoning and positive reasoning. The purpose of the framework described here, therefore, is to provide an operationalisation of these three processes and their associated products. The first part of the framework comprises a general model of argumentation, which is described below.

In his book, The Uses of Argument, Toulmin (1958) devised a general conceptual model of the functional components and structure of complex arguments.  This seminal work was based on analysis of reasoning practice (e.g. legal jurisprudence), and is one of the earliest examples of what has become the field now known as informal logic (Blair and Johnson, 1980; Kahane, 1992). Toulmin believed that the formal logics of philosophy and mathematics provided little insight into human reason, and he proposed a scheme to analyse the logical structure of everyday arguments. The Toulmin general model proposes conceptual categories which describe the component parts of an argument, and relates these conceptual categories by laying them out in a pattern which is intended to make explicit the sources of an argument’s validity.

Initially, Toulmin distinguishes between: the claim (C) or conclusion whose merits the argument is seeking to establish; the data (D) which are appealed to as the foundation for the claim; and the warrants (W) which are the laws, rules, principles, premises and so on that authorise the movement between the data and the claim. Warrants show that, if one takes some data as a starting point, the step to the claim or conclusion is an appropriate or legitimate one. A warrant is the rationale for the movement from data to claim, and can be thought of as the because part of an argument. These three elements comprise the skeleton of a pattern for analysing arguments, as shown in Figure 4:

Figure 4: Basic Structure for Analysing Arguments

Warrants, however, are of different kinds, and may confer different degrees of support to the conclusions they justify. Some warrants authorise one to accept a claim unequivocally (given the appropriate data). Others authorise one to make the step from data to conclusion either tentatively, or else subject to conditions, exceptions or qualifications. Toulmin, therefore, adds two further concepts to the framework: a qualifier (Q) that indicates the degree of support which the data confer on the claim by virtue of the warrant; and a rebuttal (R) that indicates the circumstances under which the authority of the warrant would have to be set aside. Qualifiers and rebuttals help assess the plausibility of an argument. Figure 5 illustrates how these two concepts contribute to the pattern of the argument.

Figure 5: Qualifiers and Rebuttals

In addition to the question of whether, or under what conditions, a particular warrant is applicable in a specific case, it is often necessary to ask why in general a warrant should be accepted as having authority. That is, a warrant always rests on certain other assurances, called backing (B), without which the warrants do not possess authority. The backing needed to establish warrants may exhibit great variability, but may include assumptions based on theory, practical experience, beliefs, values, and so on. The backing is included into the argument pattern as shown in Figure 6.

Figure 6: Backing

Figure 7 shows Toulmin’s original example instantiation of the complete form of the model.

Figure 7: Example Instantiation of the Toulmin Model

The general model also allows us to consider multiple chains of argumentation. For example, that data of one argument may be the claim of a previous argument or arguments (see Toulmin et al, 1984). Conversely, challenging a particular claim, may lead to more general challenges to the legitimacy of a whole range of arguments, where the claim of one argument may be the data, warrant or backing of another.

The following sections consider the characteristics of the Toulmin model with respect to the component parts of the dialectic process.

3.3.1.1  Advantages of, and Alternatives to, the Toulmin Model

The Toulmin model has a number of characteristics that make it particularly suitable with respect to the aims of this paper (i.e. supporting the accommodation of alternative paradigms, and hence supporting dependent multidisciplinary practice).

For example, Toulmin has emphasised in his model that reasoning is often contingent upon the knowledge and beliefs of those conducting the argument, and that our ability to reason is not based on infallible knowledge leading deductively to certain conclusions. Moreover, he suggests that warrants (that is, our assumptions about relationships among ‘facts’) come from our past experience and beliefs. In addition, and importantly for the dialectic process, the Toulmin model also provides a minimum set of categories for identifying what it is possible to dispute and how (Newman and Marshall, 1990).

There have been a number of other approaches that have used models of argumentation to analyse, understand, and improve problem solving activities (and design in particular). Many of these approaches have originated in the design rationale (DR) community (e.g. MacLean, et al, 1989; Lee and Lai, 1991a, 1991b). Most of these approaches, however, concentrate on representation/notations to support argumentation (see Buckingham Shum et al, 1997) and do not address the structure of argumentative reasoning per se. One notable exception is the Argument Representation Language (ARL – Smolensky et al, 1988). ARL provides a detailed, user extensible, vocabulary for describing the elements of an argument and the relations among them. The main aim of ARL is to support written scientific reasoning by providing a representation of the underlying argument structures typically found in academic papers. Because of this orientation, however, ARL is not particularly applicable in the context of system design.

As indicated above, most of the other DR-based approaches are essentially concerned with different argument representation schemes or notations. For completeness, several of these approaches are briefly outlined below.

Maclean et al (1989, 1991) have proposed the QOC (Questions, Options, and Criteria) notation as a means to represent (mainly HCI) design rationale. QOC aims to encourage designers to break their arguments down into their constituent parts and make the structure of their reasoning more explicit (Duke and Harrison, 1995). In QOC, arguments are broken down into four constituent parts: questions are used to identify key problems in the design; options are alternative answers to questions; criteria are the means by which one alternative is chosen in preference to another; and assessments are the relationships between options and criteria (e.g. supports or objects to). QOC, has been found to be useful as a medium for information storage and a language for framing the design space (Buckingham Schum, 1994). However, it does not seem to capture the (cultural) assumptions behind a design, or promote collaboration amongst designers (Karsenty, 1996), particularly when the design space is relatively well understood (Buckingham Shum et al, 1997).

The IBIS (Issue-Based Information System) notation (Kunz and Rittel, 1970) and gIBIS its modern graphical variant (Conklin and Begeman, 1989; Conklin and Burgess Yakemovic, 1991) is broadly similar to QOC, but aims to capture the design process for a single design rather than for a design in relation to its alternatives (Buckingham Schum and Hammond, 1994). In IBIS and gIBIS, designers propose issues to which they propose various answers (called positions), and then use arguments to debate the pros and cons of the proposed answers. Design using IBIS and gIBIS, then, can be characterised as a process of raising and deciding upon appropriate answers to a set of related issues (Shipman and McCall, 1997). However, from our current perspective, the limitations of IBIS and gIBIS is that there are no explicit structures for qualifying an argument or for proposing a counter argument, and it is not possible to isolate the implicit assumptions underlying arguments (Lee and Lai, 1991a).

DRL (Design Representation Language – Lee, 1990a, 1990b; Lee and Lai, 1991b) provides an extensive vocabulary for representing and managing the qualitative elements of decision-making in design, including: the alternatives being considered; evaluations of those alternatives; and the arguments and criteria behind such evaluations. As such, DRL combines a complex representation of the decision problem with a simplified model of argumentative reasoning consisting of claims, qualifications and supports-denies relations. The DRL vocabulary is intended to be extensible according to the needs of the design domain being considered (e.g. Lee, 1991 describes an extension of DRL that supports software engineering). In this sense, DRL emphasises the use of domain-specific models of argumentation, rather than with characterising arguments in terms of their generic functional structure.

In sum, then, the Toulmin model is considered to be most applicable for the purposes of this research. It comprises a useful set of distinctions, provides a basic set of categories for understanding how arguments proceed, and seems general enough to capture the basic structure of most argumentative discourse (Newman and Marshall, 1991). It also emphasises the need to explore the knowledge and assumptions (backing) that underlie particular arguments. These features are mostly absent from the other approaches described above which typically focus on design problem decomposition and its representation.

Finally, it is pertinent to note that the Toulmin model has been widely used as an argumentation scheme in a diversity of fields; including: organisational decision making (Mason and Mitroff, 1981); cognitive science (Voss et al, 1983); risk analysis (Faragó et al, 1989); public policy and law (Taylor et al, 1989); group working (Storrs, 1989); and education (Allegretti and Frederick, 1995).

In the next section, we consider some criticisms and enhancements of the Toulmin model.

3.3.2  Some Criticisms and Enhancements of the Toulmin Model

Newman and Marshall (1991) in an unpublished technical report suggest that, whilst the Toulmin model provides the minimum elements we need to understand how arguments proceed, some of these elements may be usefully extended, and that different kinds of connectivity between its elements may be possible and desirable. Their work is mostly based on the analysis of legal reasoning in two US Supreme Court cases.

In the current context, their most important criticism is that rebuttals are the most problematic and undeveloped aspect of the Toulmin model.

As discussed in Section 3.1.1, Toulmin conceptualised rebuttals as the circumstances under which the authority of the warrant would have to be set aside. Toulmin clearly sees rebuttals as being attached to the qualifier, and thus to the relationship between the data and the claim as mediated by the warrant. Newman and Marshall (1991), however, suggest that this conceptualisation is insufficient to characterise the type of legal argument which they studied, and that in particular rebuttals can sometimes attack other elements of the model. These possible different types of rebuttal are considered below.

1) Rebuttals to the data: Any of the evidence put forward in support of a claim may be questioned. Such questioning may take the form of providing evidence to the contrary, or by proposing an alternative characterisation of the evidence supplied. Such rebuttals may be concerned with any part of sometimes multiple sources of evidence.

2) Rebuttals to the claim: An argument may be rebutted by a counter-claim (with or without evidence). Counter-claims may be of two sorts: those that represent an exception or limiting case to the warrant; and those that represent a complete rejection of the warrant.

3) Rebuttals to the warrant: If too many exceptions to the warrant arise, the warrant may be attacked directly (rather than indirectly through its relationship to the data and claim). That is, the basis of the warrant may be attacked as not being relevant to the argument in hand, or sometimes its applicability to the current argument may be questioned by distinguishing the current argument from the arguments in which the warrant originally arose.

4) Rebuttals to the backing: One may attack the assumptions (that is, the theory, practical experience, beliefs, values, and so on) underlying an argument by attacking the backing. Given the paradigm analysis in Section 2, this sort of rebuttal may be particularly important with respect to operationalising the dialectic process.

These different types of rebuttals may, as suggested in Section 2.1, be the source of whole chains of counter arguments, where the rebuttal becomes the claim of a previous argument or the data for a subsequent one.

By identifying or constructing rebuttals that count against the features of an argument, we might hope to improve the quality of the argument. How, though, do we assess the quality of an argument? This question is the subject of the next section, and results in further enhancements to the Toulmin model that have been developed for the specific purposes of this research.

Finally, before leaving this section, we should mention that there have been some other criticisms of the Toulmin model from work in development of computer supported argumentation environments (such as hypertext authoring systems, for example – see Marshall, 1987; Streitz et al, 1989). Many of these criticisms, however, address the suitability of the model for computerisation, and are not of particular concern here.

3.4  Criteria for Assessing the Effect of Rebuttal on an Argument

Thus far, we have argued above that the Toulmin model is suitable as a means to identify and express the underlying features of complex arguments. This section extends this view, and proposes some criteria which can be used to evaluate arguments which have been expressed in terms of the Toulmin model. These resulting evaluative criteria constitute the second part of our framework to operationalise the dialectic process.

There are two basic ways that an argument can fail to be convincing in the light of rebuttal(s): the data, warrant and backing may be unconvincing; or the way that they are used to support the claim may be unconvincing. In the first case, it is the legitimacy of the data, warrant, and backing that is being questioned. In the second case, it is the plausibility of the claim which is being questioned.

The next sections present some criteria (after Geach, 1976; Rescher, 1976, 1977; Harman, 1986; and Govier, 1992) which can be used to asses arguments. These criteria are used to establish the legitimacy of an argument’s data, warrant and backing (hereafter collectively called the grounds for an argument), and the plausibility of its claims (as evidenced by the reasoning from the grounds to the claim).

3.4.1  Criteria for Assessing the Legitimacy of Grounds

The major criterion for assessing the legitimacy of grounds is that of acceptability. Other supporting criteria include: clarity, consistency, coherence, and certainty.

3.4.1.1  Acceptability

The grounds of an argument are acceptable if it is reasonable for those to whom the argument is addressed to believe those grounds. Acceptability implies that there is good reason to accept an argument’s grounds (even if they are not known for certain to be true), or that there is no good evidence to indicate that the grounds are false or flawed.

Some of the knowledge needed to assess any given argument will be highly specific, but there are, however, certain general conditions under which an argument maybe deemed to be acceptable.

The grounds for an argument may be acceptable if they are the output of a sub-argument that has already been shown to be cogent, or if the grounds are defended elsewhere, or if they are known a priori to be true. Grounds for an argument may also be deemed to be acceptable on the basis of a person’s testimony, particularly if that person possesses proper authority (e.g. as an established expert in an agreed field of knowledge). Such testimonies, however authoritative, may be unreliable, unsubstantiated, controversial and so on.

Arguments can be deemed as provisionally acceptable if one does not have a definite basis for considering them unacceptable. Other criteria can then be used in conjunction with acceptability to evaluate the argument.

3.4.1.2  Clarity (Explicitness)

The grounds of an argument should not be vague or ambiguous to the point where one cannot determine what sort of evidence would establish them. Grounds must be capable of being explicitly stated in specific or concrete terms (hence resolving vagueness and ambiguity by definition).

3.4.1.3  Consistency

The grounds of an argument should be consistent in the following ways: of themselves; internally in relation to other components of the argument; and externally in relation to other arguments.

3.4.1.4  Certainty

The criterion of certainty asks ‘how likely are these grounds to be true?’ Grounds should be as ‘self-evident’ and certain to be true as possible. Additionally, given that the purpose of an argument is to lead to a rational acceptance of its claims, the grounds of an argument should be more certain than its conclusion. In some cases, we make seek to verify the grounds.

3.4.1.5  Coherence

The degree to which the grounds for an argument agree with previous knowledge, experience, and beliefs. Coherence implies a mutual support, and mutual intelligibility, between the grounds for an argument (Harman, 1986).

3.4.1.6  Completeness

The grounds for an argument should be procedurally complete (that is, there should not be elements missing – e.g. data for instance); and substantially complete (that is, enough information should be provided for the purposes of the argument – e.g. more than one item of data may be required).

3.4.2  Criteria for Assessing the Plausibility of Claims

Plausibility is the strength of support that an argument’s grounds provide for its claim (Rescher, 1976, 1977), and can be assessed using two related criteria: relevance and sufficiency.

3.4.2.1  Relevance

The grounds of an argument should be relevant to its claims. That is, they should provide at least some evidence in favour of the claim being true or plausible.

We can distinguish between three types of relevance: positive relevance; negative relevance; and irrelevance. Positive relevance is where the grounds for an argument support or count in favour of one another. Negative relevance is where the grounds for an argument undermine or count against one another. Irrelevance is where the grounds neither count for or against one another, and essentially have no relationship between them.

Relevance can be assessed by use of deductive reasoning, analogy, and inductive reasoning. In deductive reasoning, the grounds, when taken together, give full logical support to the claim. If true, they prove the claim, and therefore provide good reason to believe it. In deductive reasoning, then, the truth of the claim is established from the truth of the grounds. This form of formal (and often mathematical) reasoning is least applicable to the structure of argument we have examined, since it is based on traditional models of argument (such as syllogisms) which are not used here.

The use of analogy allows the use of information from one case that may well be relevant to another. When two cases are known to be similar in a number of respects, the use of analogy provides a basis from which to infer that they may be similar in further respects also. The basic similarities of an analogy can, thus, be used to indicate relevance (but do not provide conclusive evidence of relevance).

In inductive reasoning, it is assumed that past cases are relevant to future ones and inferences are made on this basis; that is, evidence is put forward in support of some further claim. Inductive reasoning can be usefully classified into categories such as: reasoning from generalisation (that is, extrapolating from a sufficiently large and representative sample to the population as a whole); reasoning from sign (that is, reasoning from indicators which signal the presence of some referent); and reasoning from cause (that is, reasoning about the connection between conditions or events). Like analogies, and unlike deductive reasoning, inductive arguments cannot absolutely prove their claims to be true.

3.4.2.2  Sufficiency

The grounds of an argument should provide sufficient support or evidence for the claim. That is, in addition to being relevant (counting as evidence for the claim), the grounds should provide enough evidence, or enough reasons, taken together, to make it rational to accept the claim.

The relevance of grounds is distinguished from their sufficiency because though the grounds of an argument may be relevant to a claim they may not provide enough evidence to render it plausible. On the other hand, if the grounds of an argument provide sufficient evidence to make it rational to believe the claim, they will be relevant as well.

There are various sorts of sufficiency (and hence a number of ways by which grounds may be properly connected to a claim), including both formal logic (e.g. categorical and propositional logic) and informal logic (e.g. inductive support, analogy and conductive support).

For the purposes of this research, we will concentrate on conductive support and ignore formal deductive logic, inductive reasoning and reasoning by analogy. The reason for this decision, is that by adopting the Toulmin model we are structuring arguments as a number of grounds which if taken together offer support to the claim being made. That is, in conductive reasoning the grounds support the conclusion convergently, rather than in the linked way typical of formal deductive logic, inductive reasoning and reasoning by analogy. Additionally, the recognition of rebuttals is a common feature of conductive argument, where a rebuttal can be seen as being negatively relevant to the claim.

Fundamentally, evaluating the sufficiency of an argument depends on how the data, warrant and backing weigh up against the rebuttal. It is a creative task of evaluation to determine how much the rebuttal counts against the claim or whether the grounds outweigh their impact. In this sense, to assess the sufficiency of an argument, one has to construct a judgement about the relationship of the grounds to the claim in the light of the rebuttal. Such judgements must balance all considerations, and whilst it would be difficult to reduce the process to a simple formula, we can set out a logical process for making judgements:

1. Determine whether the grounds offered to support the claim are positively relevant to it (assuming they have already been found to be acceptable).

2. Determine whether any rebuttals are negatively relevant to the claim.

3. Identify any other additional rebuttals, not already identified, that are negatively relevant to the claim.

4. Reflect on whether the data, warrant and backing, taken together, outweigh the rebuttals taken together, and make a judgement as to their sufficiency in supporting the claim. Articulate reasons for that judgement.

Following this procedure helps determine whether the stated grounds provide good reason to accept the claim, and weighs up the pros and cons of an argument in a way which is a typical feature of practical decision-making (Ellman, 1995).

3.4.3  Acting on the Assessment of Legitimacy and Plausibility

Arguments that are deemed to be both legitimate and plausible may be thought of as being cogent (Govier, 1992). The concept of cogency is similar to the concept of soundness more familiar to formal logicians, but is somewhat broader in scope in that it includes arguments where the reasoning from grounds to claim is valid only in the sense of being credible rather than being logically deduced, and arguments that have plausible claims which are not necessarily known to be true.

For each argument (i.e. requirement) we construct, we must ask if its grounds are acceptable, and its claims relevant and sufficient (in that order) in the light of the rebuttal. If any one of these criteria is not satisfied, then the argument is no longer cogent in the light of the rebuttal(s).

However, to show that an argument is not cogent is to show that its claims are not well supported by the evidence put forward. However, it is important to remember that an argument is not refuted by showing that one or more of its elements is faulty. To refute it, one has to propose an independent argument whose claim contradicts the original claim.

Once we have established that an argument is not cogent, we may wish to take further actions (such as modifying the argument, constructing a new argument and so on).  Judging each argument ‘in the light of its rebuttals’ is important, since we assume that each argument is cogent in and of itself (that is, prior to facing rebuttal).

As we have seen in Section 3.4, assessing the cogency of an argument is essentially an act of judgement supported by the argument framework and criteria to establish the degree of legitimacy and plausibility it exhibits. We can further support such judgements by rating each argument, and making explicit the relationship between legitimacy and plausibility. A simple graphical technique for making explicit judgements about an argument is shown in Figure 8 (amended after Mason and Mitroff, 1981).

Figure 8: Plotting the Relative Cogency of Arguments

The graph represents a two-dimensional ordinal ranking of each argument in the light of its rebuttals, and its purpose is to make explicit the judgement concerning a given argument, and to suggest suitable courses of action according to the argument’s relative position on the graph (most importantly, which quadrant it occupies). Each of the four quadrants is considered below.

Quadrant a: high legitimacy/low plausibility

This quadrant contains arguments whose claims are placed in doubt even though the grounds for the claim are not questioned. In this case, modification of the claim is suggested.

Quadrant b: high legitimacy/high plausibility

This quadrant contains arguments whose rebuttals have provided little or no reason to doubt the original argument at all. In this case, the argument can be accepted without modification.

Quadrant c: low legitimacy/low plausibility

This quadrant contains those arguments placed in most doubt in the light of their rebuttals. As such, these arguments demand the most radical actions. There are two possibilities for action here: rejection of the original argument and its replacement with a counter-argument constructed from the rebuttal; or a synthesis of the argument and the counter-argument leading to a resolution of their opposition and the construction of a new more encompassing argument.

Quadrant d: low legitimacy/high plausibility

This quadrant contains arguments whose claims are not doubted even though the grounds for the claim are questioned. In this case, further research and analysis is necessary to improve the grounds for the argument (ultimately moving it into quadrant b).

Table 3 summarises the actions taken in response to positions in the quadrant.

Table 3: Actions Taken in Response to Positions in the Quadrant

Finally, as well as suggesting possible actions, the relative position of an argument on the graph also provides a way to prioritise arguments, and allocate available resources to future actions.

3.5  Summary

In this section, we have proposed a framework that operationalises the dialectic process and its associated products as described in Section 3.2. Specifically, the framework operationalises the three key features of the dialectic process of concern here, namely: reflection; negative reasoning; and positive reasoning. It also operationalises the three products associated with these processes, respectively: understanding; dialectic comment; and resolution. The processes, products and their operationalisations are summarised in Table 4.

Process	OperationalisedProcess	Product	Operationalised Product
Reflection	Construct Requirements Argument	Understanding	Specification of Data, Claim, and Warrant
Negative Reasoning	Critique Requirements Argument	Dialectic Comment	Specification of Rebuttals and Qualifiers
Positive Reasoning	Assess Requirements Argument and Act on this Assessment	Resolution	Statement of Cogency of Requirements Argument and Planned Actions

Table 4: Dialectic Processes, Products and their Operationalisation

The framework operationalises the dialectic processes as follows. The framework operationalises reflection via the construction of a requirements argument. It operationalises negative reasoning via the critique of the requirements argument. It operationalises positive reasoning via the assessment of the cogency of the requirements argument, and the identification of further courses of action (e.g. the acceptance, rejection or synthesis of arguments, modification of claims; or further analysis).

The framework operationalises the dialectic products as follows. It operationalises understanding via the specification of the data, warrant, background and claims of an argument. It operationalises dialectic comment via the specification of rebuttals, qualifiers, and counter arguments. It operationalises resolution via the specification of the cogency of arguments using explicit criteria, and the specification of future actions .

In Section 4, we consider how the dialectic approach (that is: the dialectic process, its associated products, and their respective operationalisations) might support the practical accommodation of paradigms.

4  Can the Dialectic Approach Support the Practical Accommodation of Paradigms?

In Section 2, we proposed a number of criteria that the dialectic approach should aim to meet in order to support the practical accommodation of paradigms, and hence overcome some of the paradigmatic constraints acting upon dependent multidisciplinary practice (see Table 1).

This section illustrates informally how the dialectic approach proposed in this paper might meet these criteria.

We examine the three main constituents of the dialectic process (i.e. reflection, negative reasoning, and positive reasoning) and their operationalisation; and its three respective products (i.e. understanding, dialectic comment, and resolution) and their operationalisation; and discuss how the dialectic approach might be said to meet the proposed criteria.

4.1  Reflection and Understanding

Reflection is essential if the practitioner is to be able to examine (and sometimes overcome) their biases about what requirements are, what they ought to be, why they are the way they are, how they can be gathered and represented and so on. This analyst based information (as well as the reasons, rationale and assumptions underlying it) must be made explicit, so that the basis of their judgement can be brought to the surface for examination (Esterson, 1970).

Essentially, the framework operationalises reflection insofar as it makes clear all the assumptions upon which the data, warrant, background and claims of a requirement are based. Moreover, reflection requires that practitioners learn to reason (in a coherent and logical fashion) from the claim to the grounds for that claim and back again.

When examining the backing to a requirement, the practitioner may examine the underlying assumptions of the technique which generated those requirements. Questions such as: “under what view of the world is this the ‘optimal’ technique to follow?; what does this technique take for granted?; what ground is it standing on?” might be asked.

By examining the data on which a requirement is based, practitioners may seek to identify the concepts they supply to practice in terms of how the data are elicited and represented. That is, a practitioner might ask: “what kind of practitioner am I?; what kind of practitioner do I need to be to use this technique?; and am I competent to collect the data required?”

By examining the warrant that justifies a requirement, the practitioner might seek to identify the benchmarks of rigour that they have used to justify a claim, and the ontology, methodology and epistemology to which they are appealing.

When examining the claims made by a requirement the practitioner might ask: “is this requirement congruent with my fundamental values and beliefs?; or, do I like what I get when I specify this requirement?”

Reflection, then, makes practitioners aware of the way that he or she has framed the requirements problem, and alerts them to the possibility of alternative ways of framing it (Schön, 1983). Practitioners must take notice of their values and norms to which they have given priority, of those that they have given less importance, and of those which they have left out altogether (Schön, 1983).

The dialectic approach generates a deeper understanding of the current assumptions that have underpinned the development of interactive system requirements, and also a deeper understanding of alternative assumptions that could be used instead. The approach may also provide an appreciation of the fact that different stakeholders conceive of requirements in different ways, and why it is important to respect such differences.

Making such assumptions explicit has the benefits of allowing judgements to be made about the efficacy of the assumptions, raises pertinent issues that might have otherwise been ignored, and also stimulates doubt (Mason and Mitroff, 1981) which is the subject of the next section which addresses negative reasoning.

In sum, relating the arguments presented above to the criteria proposed in Table 1, we might say (informally at least) that via the support it provides for reflection the dialectic approach meets criteria v), vi), ix) and x). That is, it can:

5) Help improve communication between practitioners of different paradigms (Guba and Lincoln, 1989)

6) Help practitioners understand the practical implications of alternative paradigm positions (Skrtic, 1990).

9) Bring the role of the paradigm be brought into the realm of conscious thought (Schön, 1983).

10) Support practitioners in the conscious articulation of alternative paradigm positions (Crandall, 1990)

14) Support practitioners in the management of commitment and critical reflection (Reinharz, 1981).

4.2  Negative Reasoning and Dialectic Comment

Negative reasoning serves the purpose of unearthing assumptions, testing our prejudices, and challenging our existing concepts (Posner et al, 1982). In the process of determining opposing arguments, and specifying rebuttals to requirements and their grounds, practitioners undergo a process of discovering inadequacies in their arguments by explicitly looking for conflicts between the outputs of different requirements models and techniques.

The framework operationalises negative reasoning insofar as it supports reasoning about rebuttals, qualifiers, counter requirements and their assumptions.

Specifying a rebuttal forces the practitioner to reconsider the argument and its components. Practitioners, then, must be able to determine the points at issue, and recognise alternative positions as different. The general model of argumentation supports negative reasoning by allowing the expression of different specific models. In this sense, the framework supports the detection of inconsistencies and ineffectiveness in the construction of requirements.

There is a paradox in negative reasoning, however, in that the practitioner must learn to maintain a commitment to his or her perspective (in order to best act upon it) whilst developing a readiness to abandon it. But one benefit of negative reasoning is that the complex cognitive conflict between opposing ideas is out in the open.

Negative reasoning, then, provides the practitioner with an opportunity to try to understand an alternative position and the consequences and implications of its application or adoption. During the process of negative reasoning, practitioners attempt to contradict their understanding, and (if necessary) to change their concepts and assumptions, which is the subject of the next section which addresses positive reasoning.

In sum, relating the arguments presented above to the requirements proposed in Table 3.3, we might say (informally at least) that via the support it provides for negative reasoning the dialectic approach meets criteria i), ii), iii), vi), vii), vii), ix), and x). That is, it can:

i) Support multiparadigmatic contributions to the same inquiry (Austin, 1990).

ii) Promote an appreciation of the alternative views of different paradigms (Austin, 1990).

iii) Allow each paradigm to express itself independently of the other (Burrel and Morgan)

vi) Support practitioners in the conscious articulation of alternative paradigm positions (Crandall, 1990)

vii) Support practitioners in the sequential use of paradigms (Burrel and Morgan, 1979).

viii) Support practitioners in the translation of one paradigm to another (Kuhn, 1970).

ix) Support practitioners in the enumeration of specific paradigm positions and criticisms (Reinharz, 1981).

x) Support practitioners in the management of commitment and critical reflection (Reinharz, 1981).

4.3  Positive Reasoning and Resolution

The major purpose of positive reasoning is not to justify one’s own view or to attack the weaknesses of other views, but to form a connection between them that allows their mutual exploration by all parties (Guba and Lincoln, 1989). The aim of this process is to reach a consensus decision when that is possible. When such a consensus is not possible, the process at the very least exposes and clarifies the several different views and allows the building of an agenda for further research and argumentation. The process might be thought of as successful (even when consensus is not achieved), if all views are given full consideration and new levels of information and sophistication are reached (Mason, 1969).

Positive reasoning, in contrast to negative reasoning, grasps the interconnections between concepts, not just their differences, and helps practitioners to see the larger picture. In the face of contradictions, its goal is to resolve differences of opinion and perspective by looking for a new point of view (Eemeren and Grootendorst, 1992; Yadav and Khazanchi, 1992).

The framework operationalises positive reasoning, insofar as it allows us to judge the cogency of arguments, using explicit criteria, and supports reasoning about future actions (where further action might comprise: acceptance, rejection or synthesis of arguments, modification of claims; or further analysis, for example).

Positive reasoning, then, searches for creative resolutions in the face of conflict, and as such demands a discussion-minded attitude, starting from an increased awareness of one’s own conceptual position and the opposition to it (via reflection and negative reasoning). It also demands that we be prepared to clarify and develop our current notions.

Finally, it is worth noting that not only may the requirements be reconstructed but the practitioners themselves may reconstruct their own conceptual positions (Gadamer, 1975; Solomon, 1983). For example, positive reasoning may support speculation about personal beliefs. Although beliefs cannot be immediately switched on and off at will, they are to some extent under our control and may be modified (Geach, 1976).

In sum, if the process of positive reasoning is successful, the dialectic practitioner may have built a new agenda for negotiation, and may have reconstructed the concepts with which he or she began, and in so doing have developed a more informed and sophisticated perspective.

Relating the arguments presented above to the requirements proposed in Table 1, we might say (informally at least) that via the support it provides for positive reasoning the dialectic approach meets criteria i), iii), and iv). That is, it can:

i) Support multiparadigmatic contributions to the same inquiry (Austin, 1990).

iii) Allow each paradigm to express itself independently of the other (Burrel and Morgan)

iv) Support communication between different paradigms (Guba and Lincoln, 1989)

4.4  Summary

This section has illustrated how the dialectic approach might have the potential to support the practical accommodation of paradigms which was mooted in Section 2 as being necessary in order to overcome the problems associated with dependent multidisciplinary practice by practitioner.
In the next section, we provide an illustrative application of the dialectic approach to dependent multidisciplinary practice in requirements engineering.

5.  Applying the Dialectic Approach to Dependent Multidisciplinary Practice in Requirements Engineering

To test the proposal that the dialectic approach might have the potential to support the practical accommodation of paradigms, we first have to select some appropriate paradigms/disciplines to be placed in dialectic relief. That is, we have to select some techniques to which the dialectic approach can be applied. Appropriateness can be judged, in the context of this research, by the degree of opposition and conflict between the paradigms/disciplines. That is, a dialectic approach typically requires at least two alternative perspectives which are in opposition, and each must be an adversary to the other.

5.1  Selecting Two Alternative Paradigms

To select two alternative paradigms to place into dialectic relief, we will utilise the most general (and, therefore, most widely applicable) of the three paradigm perspectives discussed in Section 2; that is, the work of Guba and Lincoln (1989, 1994). Few people in any field (apart from some postmodernists, perhaps) would object to being asked to make clear their ontology, epistemology and methodology.

Table 5 summarises the four paradigms proposed by Guba and Lincoln (1989, 1994)

	Positivism	Postpositivism	Critical  Theory	Constructivism
Ontology	Realism	Modified Realism	HistoricalRealism	Relativist
Epistemology	Objective	ModifiedObjective	Subjective	Subjective
Methodology	Manipulative	ModifiedManipulative	Transformative	Facilitative

Table 5: Basic Beliefs of Four Alternative Paradigms

As suggested in above, for the purposes of applying the dialectic approach, we should select alternative paradigms on the basis of the degree of opposition and conflict between them.

There is a clear epistemological division between positivism/postpositivism and critical theory/constructivism; i.e. a division between objectivist and the subjectivist viewpoints. These two positions differ mostly with respect to the importance granted to the subject (the knower) and the object (what can be known) in the knowing activity.

The objectivist view posits that knowledge originates from the object, and assumes that knowledge results from experience obtained through the senses. Objectivism assumes the existence of a reality that is external and independent of the knowing subject, and also assumes that this reality can be known (to varying degrees and perhaps imperfectly) through experience. The subject, playing no part in the organisation of objects uncovered by experience, has a passive role in the process of knowledge production (Landry, 1995).

The subjectivist view posits that knowledge originates from the subject, and minimises the role of the object in the process of knowledge acquisition. The perceived organised properties of known objects are seen as dependent upon the subject. The subject here has an active role in the process of knowledge production; knowledge is created not discovered by the subject. In its idealistic form, subjectivism holds that what is real is a construction in the minds of individuals.

Additionally, we can see that it is its ontological position that most differentiates constructivism from the other three paradigms, and that this position pushes the epistemological position further towards the idealistic form of subjectivism (that is, where what is real is assumed to be a construction in the minds of individuals). The relativism of its ontology means that there are multiple often conflicting constructions and all (at least potentially) are meaningful (Schwandt, 1994).

Given these arguments, we can see that there is considerable opposition between constructivism and either of the two conventional paradigms (that is, positivism or postpositivism) whose epistemological positions are broadly similar (objectivism remaining a regulatory ideal even in postpositivism).  However, for the purposes of this paper we will select postpositivism as the paradigm to place in opposition to constructivism, since it probably reflects the position most representative of many techniques in current requirements engineering practice (excepting perhaps some forms of formal systems analysis – e.g. Greenspan, 1984).

5.2  Selecting Two Alternative Disciplines

Having selected our alternative paradigms, we next need to select disciplines which can contribute to the field of requirements engineering, and which include techniques which are representative of these paradigms.

Obviously, a discipline (such as sociology, say) may include theories and techniques that belong to more than one paradigm (e.g. the Chicago School and the Frankfurt School representing positivism and critical theory respectively). Because of this, we might select disciplines on the basis of their general contribution to requirements engineering, and then characterise selected techniques in terms of their specific paradigm features to ensure dialectic opposition.

For the purposes of this paper, we will select the disciplines of human factors (HF) and sociology, both of which are frequently used in interactive systems development in general and requirements engineering in particular (e.g. Macaulay, 1996; Jirotka and Goguen, 1994).

5.3  Selecting Two Alternative Requirements Engineering Techniques

Having selected our two disciplines, it is necessary to select two techniques from within these disciplines which can be said to represent the two alternative paradigms of post-positivism and constructivism. The two selected techniques are the MUSE method for usability engineering (Lim and Long, 1994) and Grounded Theory method (Glaser and Strauss, 1967; Strauss and Corbin, 1990).

Neither of these two techniques could be said to be prominent in requirements engineering. Both techniques, however, have a number of features that make them suitable vehicles for constructing requirements.

For example, the MUSE method includes an Information Elicitation and Analaysis phase which explores existing user tasks and results in the production of a Statement of User Needs. MUSE has also been used previously in the early phases of the system development process for domain modelling (Stork and Long, 1997), design problem analysis (Stork and Long, 1994), and user requirements analysis (Stork et al, 1995). With respect to the characterisation of RE proposed in Section 1, MUSE is used here to: construct an understanding of the requirements problem via the construction of task descriptions and task models of both the existing and the target worksystems; and to construct a specification of the requirements for a software product via the construction of task descriptions and task models to support the conceptual design of the target worksystem. MUSE represents the outputs of each of these processes using semi-formal notations such as structured diagrams.

The Grounded Theory method is also considered to be suitable as a requirements engineering technique, insofar as, in common with the ethonographic approach, it can be used to construct a rich description of the everyday realities of the work situation. The Grounded Theory method has also been used previously in knowledge elicitation (Pidgeon et al, 1991). With respect to the characterisation of RE proposed in Chapter 1, the Grounded Theory method is used here to: construct an understanding of the requirements problem via the construction of a rich description of work practices and organisational/social culture in the workplace; and to construct a specification of the requirements for a software product insofar as the construction of analytic categories can be used to develop an understanding of potential requirements. The Grounded Theory method represents the outputs of each of these processes using qualitative, detailed, natural language descriptions.

Table 6 contrasts the underlying paradigm positions of the two techniques.

	MUSE (Postpositivist Paradigm)	Grounded Theory Method(Constructivist Paradigm)
Ontology	Assumes an external reality characterised as work performed by systems.	Assumes multiple internal realities that are products of  human intellects.
Epistemology	Concerned with generalisable knowledge associated with objectively observable behaviours.	Concerned with reconstruction of subjective knowledge of those investigated.
Methodology	Verification of specifications via testing.	Understanding of previously held constructions.

Table 6: Basic Beliefs of MUSE and Grounded Theory Method as Paradigm Representatives

During the PhD research process, these two techniques were used to generate requirements for an accident and emergency healthcare information system, and the dialectic process was applied to the requirements generated by the methods. This application seeks to illustrate how a dialectic approach can support dependent multidisciplinary practice by practitioner in the field of requirements engineering by supporting the practical accommodation of paradigms.

In the next section, we will briefly illustrate the use of the framework for argumentation with sampled outputs of the two selected techniques (in the domain of emergency healthcare). The aim is to provide a preliminary illustration of how the dialectic approach (via its operationalisation) might support reasoning about the mutual influence of alternative requirements engineering techniques (thereby supporting dependent multidisciplinary practice by practitioner).

5.4  Preliminary Illustration of the Dialectic Approach (via its Operationalisation)

The following example illustrates the use the framework for argumentation that operationalises the dialectic process and its associated products.

Figure 9 illustrates a simplified argument about a requirement in an A&E worksystem.

In this illustrative example, the claim which is to be made is the requirement for the redesign of the A&E worksystem, the data are that which have been gathered via the application of MUSE, the warrant offers support for the interpretation of that data by appealing to the literature, the backing appeals to published statements of intent, the qualifier assigns a probability of need to the suggested requirement, and the rebuttal questions the validity of the requirement by proposing a related piece of data (obtained as an output of the Grounded Theory method) that appears to oppose the claim.

Figure 9: Illustrative Requirements Argument for an A&E Worksystem

Having constructed the requirements argument (using outputs from the application of the MUSE method) and critiqued it (using outputs from application of the Grounded Theory method), we need to assess the legitimacy and plausibility of the argument in the light of the rebuttal, and then decide on an appropriate course of action.

With respect to the legitimacy of the argument, the grounds for the argument are acceptable. There is no reason to believe that the data, warrant or backing are incorrect (though no explicit evidence from the MUSE analysis is provided). However, there are a number of concerns with respect to the plausibility of the argument. First, the relevance of the backing to the warrant is questionable. The warrant argues that high user costs were identified for many of the tasks associated with the creation of specifications, while the backing argues that doctors are disillusioned with management emphasis on administration. The relationship between the warrant and the backing, then, requires inferences to be made about the relationship between administrative tasks and the tasks necessary to create an investigation specification; which does not preclude the conclusion that the backing may be irrelevant to the warrant.

As for the criterion of sufficiency, the warrant is not sufficient, insofar as we do not necessarily have good reason to believe that writing a single digital specification will reduce user costs. Similarly, the backing is not sufficient insofar as we do not have good reason to believe that creating an investigation specification is viewed by the doctors as an administrative task. The rebuttal implies that doctors see the computer system as supporting the nursing activities rather than their own, but we do not know under what conditions the claim holds against the rebuttal. In general, then, the warrant and the backing do not hold up against the rebuttal, but the rebuttal itself is not strong enough to reject the argument out of hand.

Figure 10 shows the assessment of this argument after rating its legitimacy and plausibility.

Figure 10: Plotting the Relative Cogency of Arguments

The argument is judged to be in quadrant a) which means that its claim is in doubt, even though its grounds are not seriously threatened. In this case, the claim needs serious modification, since its plausibility is judged to be quite low. In this case, further research and analysis is necessary to improve the claim and to establish under what conditions that rebuttal holds. Further research might even push this argument into quadrant c (where it may be rejected or synthesised in the search for a resolution). This requirement certainly requires a reasonable amount of extra effort before it can be accepted.

For the sake of argument, let us assume that further research refines the view that doctors are unwilling to use the computer system, and indicates that doctors are in fact unwilling to input data to the computer (particularly if this involves the use of a standard keyboard). Conversely, they are happy to take outputs from the computer providing that they have easy access to the information they need. In this case, the rebuttal is strengthened with respect to the creation of investigation specifications, leading us to either reject the requirement as it stands, or look for a synthesis which changes the relationship of data input to the creation of single digital specifications. Here, for example, we might look at the role of nurses in supporting doctors in providing digital specifications, or we might look for ways to encourage doctors to change their perceptions of the data input task (by linking data input to the provision of data output, for example). These suggestions are not pursued further here (since the analysis is sufficient for illustrative purposes).

6  Conclusion

This paper has attempted to develop and illustrate the use of an approach which supports dependent multidisciplinary practice in requirements engineering. To summarise, we have argued that multidisciplinary requirements engineering practice is ineffective, and identified some specific problems for dependent multidisciplinary practice. We have also suggested that the incommensurability of conflicting paradigms is an underlying cause of the problems in dependent multidisciplinary practice, and proposed a number of criteria for support to overcome such problems. A form of methodological support, which it is claimed may help overcome some of the problems associated with dependent multidisciplinary practice in RE, was developed. This methodological support takes the form of a dialectic process, and its associated products, which was conceptualised and then operationalised. The operationalisation of the dialectic process was applied to requirements generated by the use of two alternative techniques (MUSE and Grounded Theory method), and this application illustrated the methodological support it offers to multidisciplinary practice.

This paper has provided a ‘first pass’ development of a dialectic approach that can support multidisciplinary practice by practitioner in requirements engineering by supporting the practical accommodation of paradigms.

It is the author’s belief that the practical success of interactive systems is dependent upon successful multidisciplinary practice, and that we should strive to develop techniques that might make practitioners more reflective about their own beliefs, and be more willing to be as critical about their own beliefs and opinions as they are about the arguments that challenge them. Such techniques might make significant impact on the quality of requirements engineering practice and ultimately on the interactive systems that we build.

It is beyond the scope of this paper to provide a detailed description of the dialectic approach in action, and readers are pointed to the original PhD research thesis (Denley, 1999) where the author attempts to illustrate this vision in the context of the re-design of an Accident and Emergency healthcare information system.

7  References

Allegretti, C., and Frederick, J. (1995). A Model for Thinking Critically About Ethical Issues. Teaching of Psychology, 22, (1), pp 46-48.

Austin, A. (1990). Discussion on Accommodation. In: E. Guba (Ed). The Paradigm Dialog. Sage: Newbury Park, California.

Barbour, I. (1980). Paradigms in Science and Religion. In: G. Gutting (Ed). Paradigms and Revolutions. University of Notre Dame Press: South Bend, Indiana.

Bhaskar, R. (1993). Dialectic: The Pulse of Freedom. Verso: London.

Bhaskar, R. (1994). Plato etc. Verso: London.

Blair, J., and Johnson, R. (1980). The Recent Development of Informal Logic. In: J. Blair and R. Johnson (Eds). Informal Logic: The First International Symposium. Edgepress: California.

Buckingham Shum, S. (1994). Design Rationale as Argumentation: What, Why, When and Who? Design Rationale Workshop, CHI‘94, 24-25 April, 1994., Boston , Mass.

Buckingham-Shum, S., MacLean, A., Bellotti, V., and Hammond, N. (1997). Graphical Argumentation and Design Cognition. Human-Computer Interaction, 12, (3), pp 267-300.

Burrel, G., and Morgan, G. (1979). Sociological Paradigms and Organisational Analysis. Heinemann.

Checkland, P.B. (1981). Systems Thinking, Systems Practice. John Wiley.

Chua, W. (1986). Radical Developments in Accounting Thought. Accounting Rev, 61, (4), pp 601-632.

Churchman, C.W. (1971). The Design of Inquiring Systems. Basic Books: New York.

Conklin, J., and Begeman, M. (1989). gIBIS: A Tool for All Reasons. Journal of the American Society for Information Science, 40, pp 200-213.

Conklin, J., and Burgess Yakemovic, KC. (1991). A Process-Oriented Approach to Design Rationale. Human-Computer Interaction, 6, (3&4), pp 357-391.

Cosier, R. (1978). The Effects of Three Potential Aids for Making Strategic Decisions on Prediction Accuracy. Organisational Behaviour and Human Performance, 22, pp 295-306.

Crandall, D. (1990). Peering at Paradigms Through the Prism of Practice Improvement. In: E. Guba (Ed). The Paradigm Dialog. Sage: Newbury Park, California.

Denley, I. (1999). Dialectic Approach to Multidisciplinary Practice in Requirements Engineering, PhD Thesis, University College London, London, UK.

Duke, D., and Harrison, M. (1995). From Formal Models to Formal Methods. In R. Taylor and J. Coutaz. (Eds). Software Engineering and Human-Computer Interaction, pp 159-173. Springer Verlag.

Eemeren, F., Grootendorst, R. (1992). Argumentation, Communication and Fallacies: A Pragma-Dialectical Perspective. Lawrence Erlbaum: New Jersey.

Ellman, J. (1995). An Application of Case Based Reasoning to Object-Oriented Database Retrieval. IEE Colloquium on Case Based Reasoning: Prospects For Applications. Institute of Electrical Engineers: London.

Esterson, A. (1972). The Leaves of Spring: A Study in the Dialectics of Madness. Penguin: Middlesex.

Faragó, K., Vári, A., and Vecsenyi, J. (1989). Not In My Town: Conflicting Views on the Siting of a Hazardous Waste Incinerator. Risk Analysis, 9, (4), pp 463 – 471.

Firestone, W. (1990). Accommodation: Towards a Paradigm-Praxis Dialectic. In: E. Guba (Ed). The Paradigm Dialog. Sage: Newbury Park, California.

Forster, M. (1993). Hegel’s Dialectic Method. In: F. Beiser (Ed). The Cambridge Companion to Hegel. Cambridge University Press: England.

Gadamer, H-G. (1975). Truth and Method. Sheed and Ward: London.

Gasson, S. (1995). User Involvement in Decision Making in Information Systems Development. Proceedings of IRIS 18, Aarhus, Denmark, August 1995.

Geach, P. (1976). Reason and Argument. Oxford University Press.

Glaser, B., and Strauss, A. (1967). The Discovery of Grounded Theory: Strategies for Qualitative Research. Aldine: Chicago.

Govier, T. (1992). A Practical Study of Argument. Wadsworth Publishing: Belmont, CA.

Greenspan, S. (1984). Requirements Modeling: A Knowledge Representation Approach to Software Requirements Definition. Technical Report CSRG-135. University of Toronto.

Guba, E.G. and and Lincoln, Y.S. (1989). Fourth Generation Evaluation. Sage: Newbury Park, California.

Guba, E.G. and and Lincoln, Y.S. (1994). Competing Paradigms in Qualitative Research. In: N. Denzin and Y. Lincoln (Eds). The Handbook of Qualitative Research. Sage: Newbury Park, California

Harman, G. (1986). Change in View: Principles of Reasoning. MIT Press: Cambridge, Massachusetts.

Hegel, G.W.F. (1807). The Phenomenology of Spirit.

Hegel, G.W.F. (1812-1816) The Science of Logic.

Hodgson, P. (1993). Hegel’s Approach to Religion: The Dialectic of Speculation and Phenomenology. In: R. Stern (Ed). G.W.F. Hegel: Critical Assessments. Routledge: London.

Hughes, J., Randall, D., and Shapiro, D. (1993). From Ethnographic Record to System Design: Some Experiences from the Field. Computer Supported Cooperative Work, 1, (3), pp 123-141.

Jackson, N., and Carter, P. (1991). In Defence of Paradigm Incommensurability. Organisation Studies, 12, (1), pp 109-127.

Jirotka, M., and Goguen, J. (Eds). (1994). Requirements Engineering: Social and Technical Issues. Academic Press.

Kahane, H. (1992). Logic and Contemporary Rhetoric: The Use of Reason in Everyday Life. 6th Edition. Wadsworth: Belmont, CA.

Karsenty, L. (1996). An Empirical Evaluation of Design Rationale. CHI ‘96, pp 150-156, April 13-18, Vancouver, Canada.

Kitchener, R. (1986). Piaget’s Theory of Knowledge. Yale University Press: New Haven.

Kuhn, T.S. (1962, 1970). The Structure of Scientific Revolutions. The University of Chicago Press.

Kuhn, T.S. (1970b). Reflections on My Critics. In: I. Lakatos and A. Musgrave (Eds). Criticism and the Growth of Knowledge. pp 59-90. Cambridge University Press.

Kuhn, T.S. (1974). Second Thoughts on Paradigms. In: F. Suppe (Ed). The Structure of Scientific Theories. pp 459-482. University of Illinois Press.

Kunz, W. and Rittel, H. (1970). Issues as Elements of Information Systems. Technical Report No S-78-2. University of Stuttgart: Stuttgart.

Landry, M. (1995). A Note on the Concept of Problem. Organization Studies, 16, (2), pp 315-343.

LeCompte, M. (1990). Emergent Paradigms: How New? How Necessary? In: E. Guba (Ed). The Paradigm Dialog. Sage: Newbury Park, California.

Lee, J. (1990a). SIBYL: A Qualitative Decision Managment System. In: P.H. Winston and S. Shellard (Eds). Articficial Intelligence at MIT: Expanding Frontiers. MIT Press: Cambridge, MA.

Lee, J. (1990b). SIBYL: A Tool for Managing Group Decision Rationale. Proceedings of the Conference on Computer Supported Cooperative Work, pp 77-92. ACM: New York.

Lee, J., and Lai, K-Y. (1991a). A Comparative Analysis of Design Rationale Representations. Massachusetts Institute of Technology, Centre for Coordination Science, Technical Report, #121.

Lee, J., and Lai, K-Y. (1991b). What’s in Design Rationale? Human-Computer Interaction, 6, (1&2), pp 251-280.

Lim, K.Y. and Long, J.B. (1994). The MUSE Method for Usability Engineering. Cambridge University Press.

Lincoln, Y. (1990). The Making of a Constructivist. In: E. Guba (Ed). The Paradigm Dialog. Sage: Newbury Park, California.

Long, J.B. & Dowell, J. (1989) Conceptions for the Discipline of HCI: Craft, Applied Science and Engineering.  In: A. Sutcliffe & L. Macaulay (Eds). Proceedings of the Fifth Conference of the BCS HCI SIG, Nottingham, England, 5-8 September 1989, Cambridge: Cambridge University Press.  pp. 9-32.

Macaulay, L. (1996). Requirements Engineering. Springer-Verlag: London.

MacLean, A., Young, R., and Moran, T. (1989). Design Rationale: The Argument Behind the Artefact. Proceedings of CHI ‘89: Human Factors in Computing Systems, pp 247-252. ACM: New York.

MacLean, A., Young, R., Bellotti, V., and Moran, T. (1991). Questions, Options and Criteria: Elements of Design Space Analysis. Human-Computer Interaction, 6, (3&4), pp 201-250.

Marshall, C. (1987). Exploring Representation Problems Using Hypertext. In: Proceedings of Hypertext ‘87, pp 253-268.

Mason, R. (1969). A Dialectical Approach to Strategic Planning. Management Science, 15, (8), pp 403 – 414.

Mason, R., and Mitroff, I. (1981). Challenging Strategic Planning Assumptions. Wiley: New York.

Mitroff, I. (1971). A Communication Model of Dialectical Inquiring Systems – A Strategy for Strategic Planning. Management Science, 17, (10), pp 634 – 648.

Mitroff, I., Emshoff, J., and Kilmann, R. (1979). Assumptional Analysis: A Methodology for Strategic Problem Solving. Management Science, 25, pp 583-593.

Newman, S., and Marshall, C. (1991). Pushing Toulmin Too Far: Learning from an Argument Representation Scheme. XEROX PARC Technical Report SSL 92-45, 1991.

Pidgeon, N., Turner, B., Blockley,D. (1991). The Use of Grounded Theory for Conceptual Analysis in Knowledge Elicitation. International Journal of Man-Machine Studies, 35, (2), pp 151-173.

Posner, G., Strike, K., Hewson, P., and Gertzog, W. (1982). Accommodation of Scientific Conception: Towards a Theory of Conceptual Change. Science Education, 66, (2), pp 211-227.

Reinharz, S. (1981). Implementing New Paradigm Research: A Model for Training and Practice. In: P. Reason and John Rowan (Eds). Human Inquiry: A Sourcebook of New Paradigm Research. Wiley: Chichester. pp 415-436.

Rescher, N. (1976). Plausible Reasoning. Van Gorcum: Amsterdam.

Rescher, N. (1977). Dialectics: A Controversy Oriented Approach to the Theory of Knowledge. State University of New York Press: New York.

Sankey, H. (1994). The Incommensurability Thesis. Avebury: Aldershot.

Scaife, M., Curtis, E., and Hill, C. (1994). Interdisciplinary Collaboration: A Case Study of Software Development for Fashion Designers. Interacting with Computers, 6 (4), pp 395-410.

Schon, D.A. (1983). The Reflective Practitioner: How Professionals Think in Action. Basic Books: New York.

Schwandt, T. (1994). Constructivist, Interpretivist Approaches to Human Inquiry. In: N. Denzin and Y. Lincoln (Eds). The Handbook of Qualitative Research. Sage: Newbury Park, California.

Shipman, F., and McCall, R. (1997). Integrating Different Perspectives on Design Rationale: Supporting the Emergence of Design Rationale from Design Communication.Unpublished paper, http://www.csdl.tamu.edu/~shipman/ aiedam/

Skrtic,T. (1990). Social Accommodation: Towards a Dialogical Discourse in Educational Inquiry. In: E. Guba (Ed). The Paradigm Dialog. Sage: Newbury Park, California.

Smolensky, P., Fox, B., King, R., and Lewis, C. (1988). Computer-Aided Reasoning Discourse Or, How To Argue with a Computer. In: R. Guindon (Ed). Cognitive Science and Its Application to Human-Computer Interaction, pp 109-162. Ablex: Norwood, NJ.

Solomon, R. (1983). In the Spirit of Hegel. Oxford University Press: Oxford.

Sommerville, I., Rodden, T., Sawyer, P., Bentley, R., and Twidale, M. (1993). Integrating Ethonography into the Requirements Engineering Process. In: Proceedings of the IEEE International Symposium on Requirements Engineering, January 4-6, 1993, San Diego, California, pp 165-173. IEEE Computer Society Press: California.

Sommerville, I., Sawyer, P., and Viller, S. (1998). Viewpoints for Requirements Elicitation: A Practical Approach. In: Third International Conference on Requirements Engineering, April 6-10, 1998, Colorado Springs, Colorado, pp 74-81. IEEE Computer Society Press: California.

Stamper, R. (1994). Social Norms in Requirements Analysis – An Outline of MEASUR. In: M. Jirotka and J. Goguen (Eds). (1994). Requirements Engineering: Social and Technical Issues. Academic Press.

Stork A. and Long J.B. (1994).  A Planning and Control Design Problem in the Home:  Rationale and a Case Study.  In: K. Bjerg and K. Borreby (Eds).  Proc. International Working Conference on Home-Oriented Informatics, Telematics and Automation, Amager, Denmark, 27 June-1 July 1994, Copenhagen, Denmark: University of Copenhagen.  pp. 419-428.

Stork A., Middlemass, J. and Long J.B. (1995). Applying a Structured Method for Usability Engineering to Domestic Energy Management User Requirements: a Successful Case Study. In: People and Computers X: Proceedings of HCI ’95. M. Kirby, A.Dix, and J. Finlay (Eds). Cambridge University Press.

Stork A. and Long J.B. (1997). Structured Methods for Human Factors Research and Development. In: G. Salvendy, M. Smith, and R. Kubek (Eds). HCI International ‘97, 25-29 August 1997, San Francisco, USA, pp 531-534. Elsevier: Amsterdam.

Storrs, G. (1989). Group Working in the DHSS Large Demonstrator Project. Technical Report, Logica, Cambridge Ltd, Cambridge, England.

Strauss, A., and Corbin, J. (1990). Basics of Qualitative Research: Grounded Theory Procedures and Techniques. Sage: Newbury Park, CA.

Streitz, N., Hannemann, J., and Thüring, M. (1989). From Ideas and Arguments to Hyperdocuments: Travelling Through Activity Spaces. In: Proceedings of Hypertext ‘89, pp 343-364.

Sutcliffe, A.G. (1997). A Technique Combination Approach to Requirements Engineering. In: Proceedings of the Third IEEE International Symposium on Requirements Engineering, January 6-10, 1997, Annapolis, Maryland, pp 65-74. IEEE Computer Society Press: California.

Taylor, A., Knox, M., and Weaver, J. (1989). Policy Argumentation Support. Technical Report, Policy and Legal Informatics Group, University of Lancaster, Lancaster, England.

Toulmin, S. (1958). The Uses of Argument. Cambridge University Press: England.

Toulmin, S., Rieke, R., and Janik, A. (1984). An Introduction to Reasoning (2nd Edition). Macmillan: New York.

Voss, J.F., Tyler, S.W., and Yengo, L.A. (1983). Individual Differences in the Solving of Social Science Problems. Individual Differences in Cognition, Vol 1, pp 205-233.

Weaver, G., and Gioia, D. (1995). Paradigms Lost: Imcommensurability vs Structurationist Inquiry. Organzation Studies, 15, (4), pp 565 -589.

Yadav, S., and Khazanchi, D. (1992). Subjective Understanding in Strategic Decision Making. Decision Support Systems, 8, pp 55-71.