Phil Turner, Susan Turner, and Julie Horton
Department of Computing and Mathematics, University of Northumbria
This paper discusses how the concepts of activity theory can be exploited to support the early stages of systems analysis and design. The work involves the capture of contextual information using structured scenarios in a format derived from activity theory and their use in the early stages of the analysis and design of a workflow system.
For some years now, sociologically and psychologically minded researchers have suggested that activity theory may have a role to play in systems development. (To give an extremely brief explanation, activity theory derives from the Soviet school of psychology, principally the work of Vygotsky, Leont’ev and Luria. The theory conceptualises all purposeful human activity as the interaction of three elements: a subject interacts with the world through material or conceptual artefacts or tools, to realise an object. This triad has been extended with three further elements: the community in which the activity is undertaken, the rules, norms or praxis governing the activity and the division of labour between subjects (Engestrom, 1987). The elements are often represented as nodes of a triangular diagram. Problems and conflicts within and between these nodes are termed contradictions, and are considered motivators for development. Activities are realised through lower level actions and operations.) Much recent work on activity theory has focussed on the modelling of learning (cf the zone of proximal development) and of work.
The appeal of the theory for those interested in human-centred systems design is clear: there is an apparently neat integration of the use of artefacts with factors normally regarded as contextual. Moreover, recognition of collaboration is central to the theory. However, while studies exist which use activity theory to model work in the context of systems design (e.g. Kuuti & Avronen, 1992; Christiansen, 1996), very little work has been done to operationalise the concepts so that they are readily usable in practical applications. Exceptions are the work of Bardram (1997, 1998) on the SAIK project and Kaptelinin and Nardi (1997), who have produced an activity checklist for use in design and evaluation. However, neither of these provides a straightforward route between activity models and mainstream systems design. Our approach uses activity theory firstly to structure analysis of the existing system, then to structure systematic scenarios describing current practice which are used both for communication with stakeholders and the identification of opportunities for change. Scenarios have been shown to be useful for a multiplicity of purposes (Carroll, 1995) and are increasingly accepted as tools by the systems development community. This route affords a natural link, via use cases, to object oriented development methods, but that is beyond the scope of this paper.
The project described in this paper develops earlier work in which activity theory is used to provide an account of collaborative design. (Turner & Turner, 1998; Turner et al., 1999) However, that work was concerned with retrospective analysis of existing activity: we now turn to the development of AT-based representations which will provide a foundation for the specification of new activities and the software tools to support them. Having taken this first step we next turned to the application of activity theory to an in-house development.
The in-house system under investigation is concerned with the enrolment of students in an academic department. It is at first sight a prime candidate for a workflow application, requiring input from students, academic staff, administrative and finance staff from several parts of the organisation, and providing outputs to parts of the University as diverse as the Library, the Students’ Union, and the service that supports academic computing. The enrolment process is largely centred round a single form which is generated, checked by students, validated by staff, used to collect data from several different sources, and finally divided so that its parts are distributed to different destinations.
After an initial scoping exercise for the project, key members of staff were selected for in-depth interviewing. The approach requires two interviews with each respondent. In the first, a range of data was collected about the individual’s contribution to the process and the context of the work done. This information was used to develop structured scenarios, which were discussed with the subject in the second session in order to verify them and elicit any further information.
The first interview with each informant aimed to examine the enrolment process from that person’s perspective. These were semi-structured interviews based on a set of questions which covered a variety of topics. Many of these were based on the "design" version of the Activity Checklist (Kaptelinin & Nardi, 1997); others were designed to elicit responsibilities and their holders (Strens & Dobson, 1994) and factors likely to affect uptake of any future system.
After the initial interviews, scenarios were developed. These consisted of a verbal description of the work, and a more visual representation, based on the activity triangle, which indicated the relevant factors at each node and any contradictions within or between nodes. The scenarios make explicit the personal activities (Turner et al., 1999) of respondents. As activities are viewed as primarily collective, it is likely that a range of individual and perhaps conflicting perceptions of the activity will exist. These can give rise to contradictions, which in turn can be the motivators for change. By modelling personal activities explicitly, we can identify and confirm differences of understanding or practice. This will also be of use when future versions of the activity are considered, as contradictions between the current and future versions may arise from these individual viewpoints.
The subsequent interviews commenced with an explanation of the format of the scenario. Scenarios relating to both the overall activity (enrolling the students) and the individual view (X enrolling students) were presented. In each case, the scenario was presented verbally in simple language, and elements relating to the nodes of the activity triangle were drawn interactively with the interviewee. Further scenarios relating to the enrolment of specific types of students were also presented. These represent actions rather than activities. Actions can be decomposed into further actions with lower-level goals; these were co-constructed with the informant as required.
Scenarios proved in general comprehensible, though some clarification was needed of the meaning of the nodes of the activity triangle. The content was confirmed, with a few corrections and additions. Co-construction of the action-level scenarios has so far seemed unproblematic. The personal activity views caused some concern when a sensitive matter concerning the allocation of workloads was highlighted. However, this was a valuable finding as the issue concerned was potentially relevant to the design of any future system. A more thorough critique will be possible in the developed version of the presentation where we will discuss the strengths and weaknesses of activity theory in this context.
Most recently the work has been extended to an external agency. This too is in the process of introducing a workflow system. The agency is involved in investigating NHS fraud and the stated aim is to increase the throughput of cases investigated from the current 150 000 per annum to 4 000 000 with only a 33% increase in staff! Add to this both national and departmental politics and the lack of user involvement in the design of the system and we are left with a set of rich and interesting problems ripe for an activity theoretic approach. We will report on the early results of this research with a particular emphasis on co-learning / co-working.
Bardram, J. E. (1997) "Plans as situated action: an activity theory approach to workflow systems," in J. A. Hughes, W. Prinz, T. Rodden and K.Schmidt (Eds.) Proceedings of the Fifth European Conference on Computer Supported Cooperative Work, Kluwer Academic Publishers, pp. 17-32.
Bardram, J. E. (1998) "Scenario-based Design of Cooperative Systems" in F. Darses and P. Zaraté (Eds.) The Proceedings of COOP ‘98, INRIA.
Carroll, J.M. (1995), "The scenario perspective on system development" in J.M. Carroll (ed.) Scenario-Based Design: Envisioning Work and Technology in Systems Development, New York: John Wiley.
Christiansen, E. (1996), "Tamed by a Rose", in B.A. Nardi (ed.), Context and consciousness: activity theory and human-computer interaction, Cambridge, Mass.: MIT Press.
Engeström, Y. (1987), Learning by expanding: an activity-theoretical approach to developmental research, Helsinki: Orienta-Konsultit.
Kaptelinin, V. and Nardi., B. (1997) "The Activity Checklist: A Tool for Representing the ‘Space’ of Context." Research Report, Department of Informatics, Umeå University 1997.
Kuuti, K. & Avronen, T. (1992), "Identifying potential CSCW applications by means of activity theory concepts: a case example", in J. Turner & R. Krauts (eds), Proceedings of CSCW ’92, ACM Press.
Strens, R. & Dobson, J. (1994), "Responsibility modelling as a technique for organisational requirements definition" Intelligent Systems Engineering, Spring 1994.
Turner, P. and Turner, S. (1998) "Structuring scenarios using activity theory", in P. Wright and R. Fields (Eds.) The Proceeding of the First International Workshop on Understanding Work and Designing Artefacts, York, Sept. 1998.
Turner, P., Turner, S., & Horton, J. (1999), "Is Activity Theory an adequate account of the use of multiple artefacts in cooperative working?", R. Fields & P. Wright (eds.) Proceedings of 2nd King’s Manor Workshop on HCI, Design for Collaboration: Communities constructing technology, York, March 1999.