Coordination and Communication Using Signs:: Studies in Organisationat Semiotics 2

Coordination and Communication
Using Signs:
Studies in Organisationat Semiotics 2
Information and Organization Design Series
series editors
Richard M . Burton
Borge Obel
INFORMATION, ORGANISATION AND TECHNOLOGY: Studies in Organi-

sational Semiotics/ edited by K. Liu, R. Clarke, P. Andersen, R. Stamper
Coordination and Communication Using Signs:
Studies in Organisational Semiotics 2
Pär J. Agerfalk
Peter B. Andersen
Peter H. Carstensen
Samuel Chong
Rodney J. Clarke
John Connolly
Owen Eriksson
Junkang Feng
Goran Goldkuhl Editors
B. van Heusden
Michael S. H. Heng
R.J. Jorna Kecheng Liu
Kecheng Liu Rodney J. Clarke
Morten Nielsen
Daniel Robichaud
Peter Bogh Andersen
Steven Verjans Ronald K. Stamper
Springer Science+Business Media, LLC

Library of Congress Cataloging-in-Publication Data
International Workshop on Organisational Semiotics (3 *: 2000 : Staffordshire University)
rc
Coordination and communication using signs / Pär J. Agerfalk... [et al.] ; editors,
Kecheng Liu...[et al].
p. cm. - (Information and organization design series) (Studies in organisational
semiotics ; 2)
"Selected papers from the 3 * International Workshop of Organisational Semiotics, 4 of
rc m
July 2000, hosted by Staffordshire University"--Pref.

Includes bibliographical references and index.
ISBN 978-1-4613-5247-1 ISBN 978-1-4615-0803-8 (eBook)
DOI 10.1007/978-1-4615-0803-8
1. Information technology-Management-Congresses. 2. Management information
systems-Congresses. 3. Communication in organization—Congresses. 4. Signs and
symbols-Congresses. 5. Semiotics—Congresses. I. Ägerfalk, Pär J. II. L i u , Kecheng,
1957- III. Title. IV. Series. V . Series: Studies in organisational semiotics ; 2
HD30.2 .1558 2002

658.4'038-dc21
Copyright © 2 0 0 2 Springer Science+Business M e d i a N e w Y o r k

Originally published by Kluwer Academic Publishers i n 2002
Softcover reprint o f the hardcover 1st edition 2002
A l l rights reserved. N o part of this work may be reproduced, stored in a retrieval system,
or transmitted in any form or by any means, electronic, mechanical, photocopying,
microfilming, recording, or otherwise, without the written permission from the Publisher,
with the exception of any material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work.
Permission for books published in Europe: permissions@wkap.nl

Permissions for books published in the United States of America: permissions@wkap.com
Printed on acid-free paper.

Contents
Contributors VIII
Editors' Preface IX
PART 1: COMMUNICATION AND PRAGMATICS
1. Accomodating Natural Language Within The Organisational Semiotic

Framework 3
JOHN CONNOLLY
2. Means of Coordination 23
PETER B. ANDERSEN, PETER H. CARSTENSEN AND
MORTEN NIELSEN
3. Intertextuality at Work: Large Scale Organisation of Workpractices 59

RODNEYJ.CLARKE
4. Actability: a Way to Understand Information Systems Pragmatics 85

GORAN GOLDKUHL AND PAR J. AGERFALK
5. Communication Quality in the Context oflnformation Systems and

Business Processes 115
OWEN ERIKSSON
6. Greimas' Semiotics and the Analysis of Organisational Action 129

DANIEL ROBICHAUD
VI Contents
PART II: ORGANISATIONAL SYSTEMS 151
7. Reconsidering the Standard: a Semiotic Model of Organisations 153

B. VAN HEUSDEN AND R.J. JORNA
8. Using Semiotic Framework to Study Social Consequences ofIT 167

MICHAEL S. H. HENG
9. Implementing IT in Production Settings 179

STEVEN VERJANS
10. A Semiotic Approach to Improve the Design Quality of Agent-Based

Information Systems 213
SAMUEL CHONG AND KECHENG LIU
11. A Semiotics Approach to Analyzing the Information Bearing Capacity

of a Data Schema 233
JUNKANG FENG
Index 255
Contributors
AGERFALK, PAR J.
Dept. ofInformatics (ESA), Orebro University, SE-701 82 Orebro, Sweden
and CMTO, Linkoping University, SE-581 82 Linkoping, Sweden,
pak@esa.oru.se
ANDERSEN, PETER B0GH

Center for Human Machine Interaction, Department of Computer Science,
Aalborg University, Denmark, pba@cs.auc.dk
CARSTENSEN, PETER H.
The IT University of Copenhagen, Denmark, carstensen@it-c.dk
CHONG, SAMUEL
School of Computing, Staffordshire University, Stafford, STl8 ODG, UK,
Y.C.Chong@staffs.ac.uk
CLARKE, RODNEY J.
Department of Information Systems, University of Wollongong, Northfields
Avenue, North Wollongong NSW 2522, Australia,
rodney_ clarke@uow.edu.au
CONNOLLY, JOHN
Department of Computer Science, Loughborough University,
Loughborough, LEI1 3TU, UK, J.H.Connolly@lboro.ac.uk
Vlll
ERIKSSON, OWEN
Dalarna University, SE-781 88, BorHinge Sweden, and VITS, oer@du.se
FENG, JUNKANG
Department of Computing and Information Systems, University of Paisley,
High Street, Paisley PAl 2BE, United Kingdom, feng-ciO@paisley.ac.uk
GOLDKUHL, GORAN
CMTO, Linkoping University, SE-581 82 Linkoping, Sweden and
Jonkoping International Business School, P.O. 1026, SE-551 11 Jonkoping,
Sweden, ggo@ida.liu.se
van HEUSDEN, B.
Castor-Project, Faculty of Management and Organization, University of
Groningen, P.O.-Box 800, NL-9700 AV Groningen, The Netherlands,
b. p. van.heusden@bdk.rug.nl
HENG, MICHAEL S. H.
School of Accounting and Information Systems, Faculty of Business and
Enterprise, University of South Australia, Adelaide SA 5000, Australia,
Michael.S.H.Heng@unisa.edu.au
JORNA, R.I.
Castor-Project, Faculty of Management and Organization, University of
Groningen, P.O.-Box 800, NL-9700 AV Groningen, The Netherlands,
r.j.j .m.jorna@bdk.rug.nl
LIU, KECHENG
Department of Computer Science, The University of Reading, Whiteknights,
Reading RG6 6AY, UK, k.liu@reading.ac.uk
NIELSEN, MORTEN
Center for Human Machine Interaction, Department of Information and
Media Science, Aarhus University, Denmark, mnielsen@imv.au.dk
ROBICHAUD, DANIEL
B-406 Bureau, Marie-Victorin House, Laboratory: A-426-4, Marie, Victorin
House, daniel.robichaud@umontreal.ca
VERJANS, STEVEN
University of Southern Denmark, Main Campus: Odense University,
Department of Organization and Management, Campusvej 55, 5230 Odense
Editors' Preface
Organisational Semiotics occupies an important niche in the research

community of human communication and information systems. It opens up
new ways of understanding the functioning of information and information
resources in organised behaviour. In recent years, a number of workshops
and conferences have provided researchers and practitioners opportunities to
discuss their theories, methods and practices and to assess the benefits and
potential of this approach. Literature in this field is much in demand but still
difficult to find, so we are pleased to offer a second volume in the mini-
series of Studies in Organisational Semiotics'.
Following the first book in the mini-series, entitled Information,
Organisation and Technology, this book contains selected papers from the
3rd International Workshop of Organisational Semiotics, 4th of July 2000,
hosted by Staffordshire University in the United Kingdom. The chapters in
this book reflect recent developments in theory and applications and
demonstrate the significance of Organisational Semiotics to information
systems, human communication and coordination, organisational analysis
and modelling. In particular, it provides a framework that accommodates
both the technical and social aspects of information systems. The mini-series
presents the frontier of the research in this area and shows how the theory
and techniques enhance the quality of work on information systems.
Participants in the workshops exchange and discuss new ideas in the
field. Because the contributors come from several different research
communities they examine each other's ideas very critically. They are
establishing a discipline for the study of information and communication in
, Details of the mini series can be found at http://www.cs.rdg.ac.uk/~sisOlkl

x
the context of organisations, whether or not they use information technology.

The workshops also serve as important platforms for anyone - educator,
researcher, practitioner or student - who is interested in bridging the gulf
between the technical and social aspects of information systems as currently
studied in organisational contexts. This gulf does not exist in the long-
established field of semiotics, which therefore has the potential to remove
many persistent misunderstandings that are holding us back in both theory
and practice.
Though this particular formal workshop took only one day, participants
received all the papers for study before hand. A participant was assigned to
each presented paper to lead a discussion. After the presentation at the
workshop, this 'discussant' contributed a critique of the paper and
suggestions for further improvement before publication.
The chapters in the book are divided into two major sections:
Communication and Pragmatics, and Organisational Systems. In Section
One, Connolly first examines the semiotic framework from the perspective
of natural language. Andersen, Carstensen and Nielsen investigate
coordination and communication using natural language and other artefacts
in a real-life setting. Clarke's paper deals with the issues at the substantive
level of information systems and business processes although from different
theoretical perspectives. Goldkuhl and Agerfalk's view of language as action
is well elaborated in their chapter. Eriksson, following a similar theorisation,
addresses the communication quality in the context of systems and business
processes. Robichaud presents Greimas' semiotics, yet another perspective,
in the analysis of organisational action.
Section Two concentrates on organisational systems, of which a
computer system may be a component. Van Heusden and lorna propose a
number of semiotic strategies and introduce another semiotic model of
organisational structures and processes. Heng attempts to tackle the wider
impacts of information technology through case studies of teaching the
semiotic framework to international students. Verjans investigates the impact
of information technology with a particular focus on sense-making in the
work floor context. The next two chapters involve the design of computer
information systems. Chong and Liu demonstrate an approach to improve
the design quality of agent-based information systems. Feng discusses the
information bearing capacity of a data scheme in the design of information
systems.
The quality of the chapters has been achieved through a number of
mechanisms. The discussants of all the papers deserve the editors' thanks for
their excellent work. Their comments and advice to the chapter authors have
contributed greatly to improving the papers. In addition to their special
thanks to the Programme Committee (listed below), the editors wish to thank
Xl
Owen Eriksson, Michael Heng, Joaquim Filipe, Vasser Ades, Samuel

Chong, for their contributions to the revision process. Zhiwu Xie, University
of Reading, assembled the material, converted it to book format and did the
copyediting and final polishing - his effort is greatly appreciated.
. Finally, a brief message about the next book on Organisational Semiotics
in this mini-series published by Kluwer Academic. It will be based on the
papers presented at the Working Conference of the International Federation
of Information Processing, Working Group 8.1 in 23-25 July 2001,
Montreal, Canada. This third book will appear soon with the title:
Organisational Semiotics: Evolving a Science of Information Systems.
Kecheng Liu, Rodney Clarke, Peter B. Andersen and Ronald Stamper

October 2001
The Programme Committee
Albert Alderson, Staffordshire University, United Kingdom

Peter B. Andersen, University of Aalborg, Denmark
Samuel Chong, University of Staffordshire, United Kingdom (Secretary)
Rodney Clarke, University ofWollongong, Australia
Jan Dietz. University of Delft, The Netherlands
Owen Eriksson, Dalarna University, Sweden
Joaquim Filipe, Setubal Institute of Technology, Portugal
Anthony Finkelstein, University College London, United Kingdom
Henk Gazendam, Groningen University and Twente University, The
Netherlands
Joseph Goguen, University of California at San Diego, USA
G6ran Goldkuhl, Link6ping University, Sweden
Marc Hatkamp, University of Twente, The Netherlands
Michael Heng, Vrije Universiteit Amsterdam, The Netherlands (now with
University of South Australia, Australia)
Mikael Lind, University of Bon1s, Sweden
Kecheng Liu, University of Staffordshire, United Kingdom (Programme
Committee Chair, now moved to The University of Reading, UK)
Kalle Lyytinen, University of Jyvaskyla, Finland
Mareike Schoop, Aachen University of Technology, Germany
Ronald Stamper, University of Twente, The Netherlands
Claude Vogel, Semio Corporation, USA.
I
COMMUNICATION AND PRAGMATICS

Chapter 1
Accomodating Natural Language Within The

Organisational Semiotic Framework
John Connolly
1. INTRODUCTION
Natural language (NL), by virtue of its central role in human

communication, plays a vital part in the functioning of organisations. Its
importance in this regard is emphasised by, among others, Stamper (1996:
365), and it is therefore essential that the organisational semiotics (OS)
framework, as presented in publications such as Stamper (1996: 351, 1997:
276) and Liu (2000), should be capable of accommodating NL within its
scope. The purpose of the present chapter is to discuss this capability. An
attempt will be made to show that the OS framework can indeed
accommodate NL, but that certain refinements are needed in order to provide
a more precise and detailed account.
This chapter is motivated by three basic considerations. The first is that
because organisations are so dependent on NL for their effective functioning,
and hence for their very survival, an understanding of NL and its role in
organisations is of manifest importance. In striving for an understanding of
NL itself, the natural starting point is the science of linguistics, which is
devoted to precisely that goal. Of course, here within the scope of a single
chapter it is possible to cover only a small area of knowledge. Nevertheless,
we shall show how certain aspects of linguistic theory may be developed in
such a way as to enhance the OS framework and thus improve its ability to
cope with NL.
The uses of NL within organisations are too numerous to list
exhaustively. They include internal and external correspondence by
4 John Connolly
electronic or paper means, the production of reports on past, present and

anticipated situations, the conduct of meetings of various kinds and the
recording of their business, the generation of the textual parts of public
relations and marketing materials, including web-sites, and so on and so
forth. This leads us on to the second consideration, which is that while
linguistics can offer a route t 1 detailed understanding ofNL itself, it cannot
be expected to provide much enlightenment in respect of the organisational
role of NL, nor on the technology which organisations employ to support
NL-based communication. For that we need to tum to a different approach,
which is offered us by OS.
It has been demonstrated by authors such as those mentioned above that
OS can be very usefully applied in the engineering of information systems.
However, the application of OS to the detailed design of information and
communication systems to support the various NL-based activities just
referred to stands in need of further exploration and refinement; and in order
for this to be done effectively, an OS framework with a finer sensitivity to
NL is desirable. That is the third consideration which motivates the present
chapter. Of course, it is not being claimed that NL is the only important
method of communication employed by organisations, nor that computerised
NL processing has yet reached an advanced stage of development.
Nevertheless, NL remains a vital factor both in the communication process
and in relation to whatever technologies are utilised to support it.
Taken together, our three motivational considerations point to the need
for an integration of linguistics and OS. It is, therefore, to this end that the
present chapter is devoted. The chapter is theoretical in nature, and can be
seen as an exercise in descriptive research. However, one of its intended
outcomes is the provision of an improved framework which can be used in
the conduct of future empirical study of NL communication within
organisations and of its support by means of computer-based tools. The
improved framework is also of potential use to system designers, as will be
pointed out during the course of the chapter.
From the point of view of the OS approach, organisations are viewed in
terms of three nested information systems. The outermost of these is the
informal social system whose mission is achieved through the use of
communicative signs under the control of responsible agents. Embedded
with this system is a formalised sub-system, which may be partially
automated to form a technical sub-sub-system. NL is inherent in the
operation of the informal social system, and is almost certain to form part of
the formal sub-system as well (for example, in the composition of standard
letters). NL (for instance, the text of such standard letters) is also likely to
be stored within technological apparatus (such as computer files), and may
in addition be handled by means of NL processing software (such as
Accomodating Natural Language Within The Organisational... 5
automatic translation programs which might be employed to help cope with

foreign correspondence). Any OS-based attempt to analyse the structure and
function of NL in any of these information systems or sub-systems will
benefit from the provision of an optimal descriptive framework, as will any
OS-based attempt at designing new systems or sub-systems of a comparable
nature. Thus, NL is highly relevant to OS, not only on a theoretical level but
also in terms of practical investigations that need to be undertaken.
.-----'
I SOCIAL WORLD
.-----'
I PRAGMATICS
I SEMANTICS
...-----'
I SYNTACTICS
1----'
I EMPIRICS
PHYSICAL WORLD
Figure 1. Stamper's Organisational Semiotics framework
2. GROUPING OF LAYERS
2.1 The Existing Framework
The OS framework recognises six levels or layers, as shown in Fig 1.

These layers are grouped into two sets, of which one relates to the signifying
function of the sign and the other to the signified. With reference to
computer-based information and communication systems, this dichotomy
serves to distinguish those levels which pertain to the technological platform
from those which relate to the human context in which the communication
takes place.
The framework just outlined can be applied successfully to the analysis
and description ofNL. For instance, consider the following sentence:
(1) Who dropped the flask?
This sentence can be analysed at all six levels, along the following lines:
(2) (a) Social world: speaker (e.g. chemistry teacher), audience (e.g.
class of students), relative status of speaker and audience, expectations (e.g.
responsible behaviour, exercise of due care), and so on.
6 John Connolly
(b) Pragmatics: speaker is opening a new section of dialogue, posing

a question with a view to eliciting a response, and so on.
(c) Semantics: 'who' denotes an unknown human agent, and so on.
(d) Syntactics: 'who' is a noun phrase, consisting of an interrogative
pronoun, functioning as subject of an independent clause, and so on.
(e) Empirics: weak aperiodic noise followed by strong periodic
resonant tone, and so on.
(f) Physical world: wave-like variations in air pressure through time.
Moreover, it is clear that (2d-f) relate to the signifier, since they concern
the form and the concrete manifestation of the utterance, while (2a-c) are
concerned with the signified, as they are to do with meaning in context. It is
(2d-f) which, moreover, pertain directly to the technological platform.
2.2 Alternative Groupings
The division of the existing framework into two sets of layers will here
be termed the signification-based grouping of levels, as it is founded on the
signifier/signified distinction. However, two alternative divisions can also
be justified. First of all, we can propose the following three-way split:
(3) (a) The core layers: syntactics and semantics.
(b) The infra layers: empirics and physical world.
(c) The ultra layers: pragmatics and social world.
The core levels are those which comprise the language code itself. This
code provides the (syntactic) units for constructing linguistic expressions and
the (semantic) units for organising the content expressed through the
structures concerned. The infra levels are so called because they are lower-
level in a scale of abstraction than the core layers. ('Infra', of course, means
'below'.) At the infra levels, language is detectable, either acoustically (in
the case of spoken language) or optically (in the case of written language).
At the core levels, in contrast, we deal with abstract phenomena which are
essentially analytical constructs rather than physical manifestations. As for
the ultra levels (given that 'ultra' means 'beyond'), these represent our
theoretical apparatus for looking outside the confines of the linguistic code
itself into the surrounding phenomena of language use in actual social
contexts. This ternary division will be termed the core-based grouping.
Both the significance-based grouping and the core-based grouping are
valid. It is therefore proposed to retain the former but also to add the latter
to the OS framework. In other words, the core-based grouping should be
superimposed upon the significance-based grouping.
.P;fiM{
SOCIAL WORLD
Ultra
Layers
Core
{I----_.....i.------L-~
Layers
Signifier
Infra
Layers
PHYSICAL WORLD
Figure 2. An enhanced version of the Organisational Semiotic framework
This proposal is reflected in Fig 2, which is a modified version of Fig 1.

In redrawing the diagram, the opportunity has been taken also to reflect
connections between non-adjacent levels in the hierarchy. Firstly, the social
and physical worlds belong together as the environment in which the
communication takes place, and are therefore shown as abutting in the
diagram. Secondly, there is a direct connection between the levels of
syntactics and pragmatics. For instance, it is well known that the order of
elements in a sentence is subject to the influence of pragmatic factors such as
emphasis or connectivity with what has gone before in the discourse; see
Connolly (1991: 10-19) for discussion and references. Compare, for
example, the following two sentences:
(4) (a) The client must have known that.
(b) That the client must have known.
These two sentences are semantically non-distinct, but differ in that the
initial placement of the object 'that' in (4b) serves to highlight it and to link it
with the preceding linguistic context. Thirdly, there is also a direct
connection between the levels of pragmatics and empirics. For instance,
consider the pronunciation of III immediately before a consonant in British
English. Typically, this pronunciation is slightly different in the speech of
Welsh people compared with English people. This distinction is not a matter
of the language code, as the Welsh III sound fulfils exactly the same function
in Welsh accents as the English III sound does in English accents. Rather, it
is a difference in the pattern of concrete manifestation of the sound in
question, and therefore, when treated from the point of view of a signifier, it
belongs at the empiric level. On the other hand, from the standpoint of what
is signified by a particular pronunciation of the III (namely the geographical
provenance of the speaker involved), this reading is a matter of pragmatics.
Individual sounds of language do not have their own semantics, but
inasmuch as their properties contribute to the overall message that is being
8 John Connolly
communicated, their interpretation forms part of the pragmatics of the

receptive processing of speech.
There is also room for a third grouping, to be superimposed upon the two
already outlined. We can distinguish between the following:
(5) (a) The communicative layers: empirics, syntactics, semantics and
pragmatics. (These together cover the substance, form, content and
interpretation of the messages conveyed by means ofNL.)
(b) The environmental layers: physical world and social world.
(These, as already stated, serve to connect NL to the context in which it is
situated.)
This is, clearly, another dichotomous split, which will be termed the
encompassment grouping, as it distinguishes the communicative process of
NL itself from the context which surrounds (or encompasses) it. For
discussion relating to context from the point of view of OS, see the
proceedings of the Second Workshop on Organisational Semiotics,
especially the papers by Stamper, Gazendam, Andersen and May, Filipe, and
Liu and Sharp. These papers also appear in Liu et al. (2000).
An additional, subordinate grouping also suggests itself. The
communicative layers consist of the two core layers of syntactics and
semantics, which act as codes, together with the levels of empirics and
pragmatics, at which the actual communicative action takes place. We thus
have the following action-based grouping of the communicative layers:
(6) (a) The core layers (syntactics and semantics).
(b) The action layers (empirics and pragmatics).
The empiric level is concerned with physically observable action (such as
movements of the vocal organs in speech) and its tangible products (for
example, sound waves), whereas the pragmatic level deals with social
interaction.
2.3 Benefit of the Alternative Groupings
The alternative groupings just proposed are useful in that they highlight
different perspectives with regard to the design of NL-based systems that
may be required by organisations. The core-based grouping encourages the
designer to distinguish between the language resources (the core layers), the
use of language (the ultra layers) and the physically perceptible aspects of
language (the infra layers). For instance, suppose that an organisation
requires a NL-based query system as an interface to a database. In this
scenario the designer will be reminded to pay adequate attention to the
coverage of the grammar and semantics (the language resources), the design
of the user-system dialogue (the use of language) and the legibility of the
screen display (the physically perceptible aspects of language). Thus, none
of these important factors would escape the designer's net. Furthermore,

consideration of the use of language will include an appreciation of the
relationship between the language itself and the context in which it is
employed. This will foster an appreciation of how the language drives the
user's task forward, and conversely, how the progress of the task influences
the language used.
Language and context, then are closely interrelated. On the other hand,
for some purposes it is useful to distinguish between them, and this is
encouraged within the perspective provided by the encompassment
grouping. This view is likely to appeal to a designer working within the
approach motivated by General Systems Theory, where the boundary
between system and environment is regarded as very important. For
instance, suppose that an organisation requires a multimedia information
system, in which most if not all of the displays will contain both NL text and
non-NL material such as images. Different design considerations apply to
the NL text and to the non-NL components which provide part of the context
of the text in question, and this contrast is highlighted by the encompassment
grouping.
As for the action-based grouping, this encourages the designer to pay
attention to the distinction between those aspects of a system that are subject
to short-term change (resulting from the action) from those which are not.
For instance, the focus on short-term change may motivate the designer to
consider whether it is necessary to provide a system with a facility for
recording the progress of a dialogue (a so-called history), in order to keep
track of the developing situation brought about by the successive
contributions made by the participants in the discourse concerned.
3. SUBDIVIDING THE LAYERS
3.1 Accepted Subdivisions
In order for the OS framework to reflect the structure of NL in more

accurate detail, it is also very helpful to subdivide most of the layers into
sub-levels. Indeed, many of these subdivisions count as standard practice in
linguistics, and can therefore be dealt with fairly briefly.
Firstly, the level of syntactics is conventionally subdivided into two:
(7) (a) An upper (more abstract) tier, to be known here as the
morphosyntactic sub-layer.
(b) A lower (less abstract) tier, which (unlike the morphosyntactic
sub-layer) is sufficiently concrete to be inherently sensitive to the distinction
between the spoken and written language. In the case of the spoken
10 John Connolly
language the lower tier is called the phonological sub-layer, while the
counterpart in the written language will here be termed the orthographical
sub-layer.
The morphosyntactic sub-layer is traditionally further sub-divided into
two parts:
(8) (a) The morphological sub-sub-layer, concerned with the internal
grammatical structure of words.
(b) The syntactic sub-sub-layer, concerned with the structure of
sentences and thus embracing the internal organisation of units such as
phrases and clauses, which are larger than the individual word.
The phonological sub-layer, too, is conventionally subdivided into two:
(9) (a) The prosodic sub-sub-layer, dealing with phenomena such as
stress and pitch, in which the fundamental unit of analysis is the syllable.
(b) The segmental sub-sub-layer, dealing with smaller units than
the syllable, including the individual consonant and vowel segments into
which syllables can be internally analysed.
By way of example, consider the following sentence:
(10) The blue flowers are looking very attractive.
The syntactic description of this sentence would concentrate on the
combination of words to form phrases, such as the noun phrase 'the blue
flowers', the combination of the phrases to form the single clause of which
the sentence consists, and the functions of the various constituents of the
sentence, for example the subject role of the noun phrase just mentioned.
The morphological description, on the other hand, would focus on word-
internal structural patterns, such as the formation of the form 'flowers' out of
the base form 'flower' plus the plural suffix's'.
In the spoken form of (10), the prosodic phonological analysis would
reveal phenomena such as the levels of accentuation of the syllables; for
instance, it is likely that 'the' would be unstressed, 'blue' would be stressed,
and the first syllable in 'flowers' would also be stressed, but the second
syllable unstressed, and so forth. The segmental phonological description of,
for example, the word 'blue' would reveal the consonant cluster Ibl + II/
followed by the vowel lui. In contrast, the orthographical description of the
word 'blue' would recognise four letters rather than the three sound segments
identified by the phonological analysis, as the spoken vowel lui is written by
the letter-pair 'ue' in this particular word.
It has been proposed by Halliday (1966) that lexis (i.e. vocabulary)
should be regarded as an additional sub-layer, alongside the
morphosyntactic. There are at least two arguments in favour of this. Firstly,
there exist many idioms (in English, at least) which consist of more than one
word, for example 'come across' in a sentence such as the following:
(11) Have you come across my notes?
To treat 'come across' here as merely a combination of verb plus

preposition (which is how it might be analysed syntactically) would be to
obscure the fact that it acts as a single unit of vocabulary, meaning 'find' or
'notice'. This is not to say that the syntactic analysis is wrong or irrelevant,
but that there is also a place for a lexical analysis in which the unitary nature
of the idiom 'come across' is afforded recognition. Secondly, besides idioms,
we encounter patterns of word-combination that are again missed in
syntactic analysis. These patterns are called collocations, and they consist in
preferences for particular vocabulary-items to occur together. For instance,
consider the following two phrases:
(12)(a) Red wine.
(b) Blue wine.
Both these· phrases are identical syntactically, consisting, as they do, of
an adjective followed by a noun. However, (12a) represents a collocation
(or habitual association), whereas (12b) clearly does not. The motivation for
the difference, in this and many other examples, may be semantic, but the
patterning itself is an observable property of the way in which sentences are
constructed, and therefore belongs at the level of syntactics.
3.2 Subdividing the Layer of Empirics
'Empirics' is not a term that is traditionally used in the description ofNL.

However, the empiric layer bears comparison with the following linguistic
concepts:
(13) (a) In respect of the spoken language
The phonetic level of analysis, which is concerned with the manifestation
of speech as a temporally co-ordinated set of movements of the vocal
apparatus, giving rise to a time-varying, audible, acoustic wave pattern.
(b) In respect of the written language
The graphetic level of analysis, which deals with the manifestation of
writing as a temporally co-ordinated set of manual movements, producing a
visible pattern on a surface.
This suggests, then, that the empiric layer, as applied to NL, should be
subdivided into to a phonetic sub-layer and a graphetic sub-layer. Such a
proposal is, indeed, to be recommended as a starting point. However, we
need to go further.
The phenomena which Stamper associates with the empiric level include
noise, entropy, channel capacity and such like. These terms are not the
common currency of phonetics, which deals with the specifically speech-
related aspects of the sound. Such characteristics include the following:
(14) (a) From the physiological point of view
12 John Connolly
The source and direction of the airflow (e.g. outward from the lungs); the
position and movement of the vocal cords (e.g. adducted together and in
vibration); the position and movements of the tongue, lips and soft palate in
relation to other parts of the vocal apparatus.
(b) From the acoustic point of view
Periodicity or aperiodicity of the wave-form, consequent upon the
position and movement of the vocal cords; presence or absence of high-
frequency friction-sound associated with narrowing of the passage between
(for instance) the lower lip and the upper teeth.
In order to accommodate both Stamper's concept of the empiric level and
the traditional concerns of phonetics, we need to propose a further
subdivision of the layer of empirics:
(15) (a) An internally-oriented sub-layer, to be known as the interior
facet.
(b) An externally-oriented sub-layer, to be known as the exterior
facet.
The interior facet of the phonetic sub-layer is concerned with phenomena
which come directly to our notice when we consider the sound produced in
speech as just sound, pure and simple, disregarding the fact that it is a
manifestation of an attempt to say anything. From this point of view, the
properties which relate to the acoustic signal, such as noise or channel
capacity, emerge as the natural focus of attention. On the other hand, the
exterior facet of the phonetic sub-layer represents a view of the sound of
speech as the material embodiment of the phonological units belonging to
the level of syntactics of the spoken language. Put differently, this exterior
facet presents us not with sound as a phenomenon in its own right, but
specifically as a vehicle for the actualisation of speech as opposed to
anything else (be it music, bus-exhaust, escaping steam, or whatever). In
this view, high-frequency noise is not considered autonomously, but is seen
as the result of producing a consonant such as lsI or Iv.
Similarly, the graphetic sub-layer has both an interior and an exterior
facet. Once again, the interior facet shows the phenomenon purely in its own
terms. The marks made on the surface are just regarded just as geometric
configurations. In this respect they may be characterised by effects such as
noise (in the form of spots, smudges and the like). However, in the exterior
facet these configurations are seen as representing letters, punctuation marks,
and so on. In other words, they are viewed as the visible manifestations of
the orthographical units which belong to the level of syntactics of the written
language. Once again, the external facet looks outwards to the level above,
which, indeed, its own level subserves.
Two further points are worth making at this stage. Firstly, the view of
one level subserving a higher level in the hierarchy implies a functional or
instrumental approach to NL, and particularly echoes the approach of Dik
(1997: 8) in his description of his theory of Functional Grammar. Secondly,
the distinction between the interior and exterior facets at the empiric level
bears out practical experience in the field of research into automatic speech
recognition (ASR); cf. Connolly et al. (1986). It is well known that
particular types of sound have particular acoustic characteristics. For
example, as already mentioned, sounds like lsi and /zl manifest a
concentration of aperiodic noise in the upper part of the acoustic spectrum.
Moreover, if the spectrum of a stretch of speech is recorded and analysed
and the results of the analysis displayed, then if one already knows what was
said, one can usually identify the key features of each sound in the sequence.
However, if one does not know what was said, then it is not at all easy to
reconstruct the spoken message. True, it may well be possible to identify
acoustic patterns such as concentrations of energy in particular regions of the
spectrum. Acoustic properties of this kind belong to the interior facet of the
empiric level, and are readily measurable as properties of sound in their own
right. However, it is very difficult to bridge the gap from the interior to the
exterior facet and to identify correctly the acoustic patterns as manifestations
of particular phonological units. (For instance, the acoustic patterns of lsi
and Izl can be hard to distinguish from one another with any confidence.)
Thus, our distinction between interior and exterior facets plays a part in
explaining why ASR is such a difficult problem.
3.3 The Interior and Exterior Facets of the Layer of

Syntactics
The distinction between interior and exterior facets turns out to be

applicable not only to the layer of empirics but to the layers of syntactics,
semantics and pragmatics as well. In other words, it is valid throughout the
communicative layers, as will now be demonstrated. The general principle
is as follows:
(16) (a) The interior facet of a given layer is concerned with the aspects
which are pertinent to that level as a more-or-Iess self-contained component
of language.
(b) The exterior facet of a given layer is concerned with the
aspects which can be regarded as manifestations (or realisations) of higher
levels in the hierarchy which is embodied in the OS framework.
At the level of syntactics, a good deal of the organisation is describable in
terms which are particular to that layer, rather than in terms of relationships
with the level of semantics which lies above it in the hierarchy. Word
14 John Connolly
classes (e.g. verb and noun) can be established on the basis of formal
criteria; for instance, verbs can be observed to occur in present and past
tense forms. Moreover, when words co-occur to form larger structures,
these have certain purely syntactic properties. For instance, consider a
simple noun phrase such as the following:
(17) A conference.
This phrase consists of the determiner 'a' and the noun 'conference'.
These two units engage in at least four syntactic relationships:
(18) (a) Co-occurrence: they appear together in a sequence.
(b) Order: the determiner precedes the noun, rather than the
reverse.
(b) Constituency: they are part of a larger whole, namely the noun
phrase.
(d) Dependency: the occurrence of the determiner is dependent on
there being a noun for it to specify.
Larger units (i.e. clauses, and sentences) are similarly describable in
terms such relationships. Furthermore, even function-related categories like
'subject' are amenable to a formal characterisation. Consider the following
example:
(19) A conference is being held here today.
The subject of this sentence is 'a conference', and as such it has the
special property of entering into a dependency relationship with the finite
verb, whereby the latter has to agree with it in terms of person and number.
In this instance, the subject is third person singular, and so is the finite verb;
it cannot be 'am' (first person) or 'are' (second person and/or plural) instead.
In phonology, too, we find phenomena which relate purely to the layer of
syntactics of which it forms part. The structure of syllables in terms of
consonants and vowels is a matter of form and has no direct relationship to
the signified. Moreover, syllabic attributes of stress and pitch form prosodic
patterns which are again amenable to formal description. For instance,
intonation patterns can be described in terms of tone units, each of which has
the formal property of containing precisely one nuclear syllable, which is
pivotal in the pitch trajectory of the tone unit concerned and normally
combines strong stress with an identifiable pitch movement. This is
obviously difficult to illustrate in a written text, but the point is simply that
prosodic patterns can be described without having to appeal to concepts
associated with other levels.
In the space available, it is not possible to offer anything remotely
approaching a complete coverage of the interior facet of the layer of
syntactics. An attempt has merely been made to illustrate the fact that such a
facet can reasonably be recognised, inasmuch as at least some phenomena at
the level of syntactics can be described, at least in part, in tenns which do

not make reference to semantics, pragmatics or social context.
Nevertheless, it is possible to view at least some aspects of the layer of
syntactics as realisations of phenomena at higher levels, and therefore to
justify the recognition of an exterior facet to the layer in question. Again, a
brief illustration will have to suffice.
Let us begin with the following example sentences:
(20) (a) Professor Smith organised a conference.
(b) A conference was organised by Professor Smith.
(c) A conference Professor Smith organised.
Let us also assume, for the sake of argument, a type of semantic analysis
whereby in (20a) 'Professor Smith' is described as the agent of action of
'organising', while 'a conference' is described as the goal of that action. The
terms 'agent' and 'goal', being semantic in nature, relate to the content of the
sentence. However, these elements of content have to receive expression in
the shape of structural units (in this instance, noun phrases). Thus, for
instance, the phrase 'Professor Smith' can be viewed from two points of
view. From the point of view of the interior facet of syntactics, it is the
subject of the sentence in which it stands. On the other hand, from the
standpoint of the exterior facet of syntactics, it is the fonnal realisation of the
semantic agent. Similarly, 'a conference' is the syntactic object, and at the
same time the fonnal expression of the semantic goal.
From the point of view of the exterior facet, the purely syntactic (i.e.
interior) properties of the expression are of secondary importance. For
instance, the semantic goal is realised by the syntactic object in (20a) but by
the syntactic subject in (20b). The precise syntactically interior status of the
element is not what really matters from the exterior point of view, provided,
of course, that it can serve as a suitable vehicle for the expression of content
(preferably in a way which is unambiguous in context).
As already noted, pragmatic as well as semantic import can be given
expression in syntactic form, and (20c) can serve as a reminder of this.
Here, the pragmatic highlighting of the goal is accomplished via the
syntactic ordering of the elements. (The pragmatic effect is to emphasise
that it was a conference that Professor Smith organised, as opposed to some
other kind of event, such as a brewery visit.) Thus, the exterior facet of the
level of syntactics is open to the layer of pragmatics as well as that of
semantics.
The phonological sub-layer of the level of syntactics also lies within the
scope of the exterior facet. For example, if a sentence like (20a) were
spoken, then the presence of a low-falling intonation pattern would indicate
that it was a statement, whereas a high-rising pattern would mark it as a
question. Depending one one's theoretical stance, the contrast concerned
16 John Connolly
might be taken as either a semantic or a pragmatic distinction. In either case,

however, the intonation pattern can be regarded as an expression of a choice
belonging to a higher level in the hierarchy.
The recognition of interior and exterior facets at the level of syntactics
probably saves us from having to recognise lexis as a separate level. Instead,
we can exploit the exterior facet of the syntactic sub-layer for the same
purpose. Idioms can be seen as multi-word expressions of unitary semantic
items, while collocations can be viewed as co-occurrence relationships
between words seen as carriers of meaning, reflecting the fact that semantics
generally plays a vital explanatory role in accounting for collocational
patterns.
3.4 The Interior and Exterior Facets of the Layers of

Semantics and Pragmatics
Semantics, like syntactics, includes certain phenomena which are

particular to that layer rather than being direct reflections of elements at a
higher level. Semantic roles like 'agent' and 'goal' are examples. So are
logical relations, such as the implication between (21 a) and (21 b) or the
incompatibility between the true statement (22a) and the false statement
(22b).
(21) (a) Pens are items of stationery.
(b) Ball-points are items of stationery.
(22) (a) Pens are items of stationery.
(b) Pens are items of footwear.
On the other hand, meanings are conveyed in order to achieve the goals
of discourse. For instance, (21a) might be spoken by a person as part of a
discourse aimed at teaching a new office worker the classification system
used when ordering supplies. Content viewed from the point of view of the
role it plays in progressing the discourse represents semantics in terms of its
exterior facet.
The actual structure of discourse, however, is a matter for the layer of
pragmatics. For instance, a conversation involving two participants will
normally be constructed on a turn-by-turn basis. Each turn constitutes what
is often called a 'move'. It pushes the conversation forward, either in the
same direction as before or in a new direction. To describe discourse in
terms of moves and other, comparable units is to present it from the point of
view of the interior facet of pragmatics. No direct reference is made to the
only higher level in the OS hierarchy, namely the social world.
Nevertheless, discourse frequently does have effects upon the social
world. For instance, if a session of staff training is successful, then the those
who attend end up knowing more than they did before the session began.
Viewing discourse in this light is to focus on its exterior facet.
It seems, then, that the concept of interior and exterior facets is, indeed
applicable to all four communicative layers of NL. The adoption of this
view serves to lend yet another pleasing symmetry to the OS framework, in
addition to those revealed by the three groupings of layers described earlier.
3.5 Benefit of Recognising Layers and Facets
NL-based systems can be complex, and so a major advantage of dividing

NL up into layers, and layers into facets, is to enable attention to be focused
upon comparatively manageable aspects of a given problem. For instance,
suppose that an organisation requires an automatic NL translation system,
which accepts spoken input and delivers spoken output. This requirement
calls for an enormously complex system, involving numerous design
decisions. Some of these decisions can, however, be identified more clearly
with the help of layers and facets. For example, should the system contain a
phonological sub-layer, or should it be designed to cope with entire words as
basic units? If it is going to contain a phonological sub-layer, then should it
include a prosodic sub-sub-Iayer, or should it manage without this difficult
component? Should it cater for the interior facet of syntax, or can it avoid
this complication and employ just the exterior facet? And so on. Such
decisions are not easy to make, but at least our NL framework helps the
designer to ask some ofthe right questions, and that of itself is a big help.
4. PARALLELS BETWEEN LAYERS
In Stamper's description of the OS framework, certain concepts are

associated with specific layers. For instance, 'pattern' is ascribed to the
empiric level. However, at least when the framework is applied to NL, we
find that some of the concepts in question apply to more than one layer.
'Pattern' is a case in point. Applied to the empiric level, this term refers
to organisation and regularity in the (acoustic or optical) material
embodiment of spoken or written language. However, organisation and
regularity are found at all levels. Among the communicative layers they are
manifested in syntactic structure (e.g. in the construction of sentences), in
semantic structure (e.g. in hierarchies such as 'device', 'peripheral', 'printer',
and so on, where 'device' is superordinate to the terms denoting its sub-
types) and in pragmatic structure (e.g. in the organisation of discourse into
moves and other units). Furthermore, various hierarchies and other patterns
can be identified both in the social world and in the physical world.
18 John Connolly
Moreover, the empiric layer is characterised by a number of statistically-

based concepts, such as variety, entropy and redundancy. However, it is
possible to attempt a comparable statistical characterisation of phenomena at
the level of syntactics. For instance, words can sometimes exhibit variety in
their composition (e.g. in the choice of initial vowel in the spoken form of
the word 'economics'); they can be more or less probable in a given context
(cf. the discussion of collocation above); and they can be employed
redundantly (e.g. in expressions like 'big, huge').
Again, the terms 'logic' and 'deduction' are assigned by Stamper to the
level of syntactics, where they can be appreciated in, for instance, the
derivation of sentences from generative grammars. However, they also
apply to the semantic layer (cf. the discussion of the interior facet of this
level, above).
These remarks demonstrate that we have to be careful how we
characterise the properties particular to any given level. On the other hand,
they also bear out an important point made by Stamper (1997: 275) that key
terms can have different senses in relation to different layers in the
hierarchy; cf. also Nauta (1972: passim).
5. NORMS
Within the OS framework, the concept of 'norms' is considered to be of

fundamental importance. The following types of norm are recognised; see
Stamper (1997: 278).
(23) (a) Cognitive: pertaining to our understanding of how the various
parts of the world relate to one another and function together.
(b) Behavioural: pertaining to our knowledge of how to act in
particular situations.
(c) Evaluative: pertaining to our discrimination between good and
bad qualities.
(d) Perceptual: pertaining to our experience of the physical and
social world, as mediated by our senses.
Moreover, each norm should be stated in terms of a conditional, 'if ...
then .. .' structure.
All four types of norm can be identified with reference to NL. An
example of each can be seen, respectively, in the following:
(24) (a) If a determiner is immediately followed by a noun, then these
two words are constituents of the same noun phrase.
(b) If a person is conversing with friends, then an informal style of
speech is appropriate.
(c) If a text is written incoherently, then it will be a poor vehicle

for communicating its intended message.
(d) Ifa person who knows English reads the sequence of letters 'p',
'e', 'n', then he or she will recognise the word 'pen'.
Of course, these are very simple examples, but they nevertheless serve to
show the applicability of the principles. They also illustrate a further point,
namely that one norm may depend on another. For instance, the perceptual
norm (24d) entails the cognitive norm that if a person knows English, then
he or she will know the word 'pen'. Without such knowledge, the
recognition of the corresponding percept will obviously not occur.
Each layer or sub-layer of NL structure has its own associated norms.
For example:
(25) (a) The phonological sub-layer encompasses norms of
pronunciation.
(b) The orthographical sub-layer encompasses norms of spelling.
(c) The morphosyntactic sub-layer encompasses norms of
grammar.
(d) The semantic layer encompasses norms of meaning, for
instance the range of senses in which each word in the language can be
employed.
(e) The pragmatic layer encompasses norms of language use, such
as politeness conventions.
It must be admitted, however, that conventional descriptions of NL do
not necessarily take the form of actual 'if ... then ... ' structures. Insofar as
such descriptions are statements of collections of norms, therefore, the OS-
style conditional formulation would have to be regarded as implicit rather
than explicit. Nevertheless, given that language operates so extensively on
the basis of convention, it is inevitable that the notion of the 'norm' should be
germane to our understanding of its functioning.
Indeed, some traditional linguistic concepts such as 'grammar' or 'genre'
might be regarded as norm-bases. Within a grammar the norms generally
take the form of rules or state-transition networks, while genres can be
described in terms of particular constellations of conventions. For instance,
the genre of a written letter is characterised by the convention that its main
part begins with 'Dear' followed by a form of address (such as 'Sir' or 'Mr
Jones'), and ends with 'Yours faithfully' or 'Yours sincerely' followed by the
sender's signature.
Modes of representation such as the rules that tend to be used to describe
grammars, or the (usually less formal) statements which describe the
conventions that characterise particular genres, are normally regarded as
quite distinct in nature from one another. However, to see these different
types of representation all as expressions of norms, as OS encourages us to
20 John Connolly
do, is to benefit from a useful unifYing concept, which applies to all six
levels.
The idea of a norm also implies that exceptions, too, can be expected, but
that such exceptions are not necessarily ill-formed. Again, this idea is
applicable to NL, where exceptions to rules are quite common, without
implying that ungrammaticality is rife.
6. CONCLUSION
Our attempt at accommodating NL within the OS framework has

involved the following:
(26) (a) Proposing two additional groupings of layers within the
framework.
(b) Clarifying the connections among the various levels.
(c) Subdividing the communicative layers beyond the degree
conventional in linguistics.
(d) Identifying various structural parallels between layers.
(e) Demonstrating the applicability, in principle, of the four
different types of norm to the description of NL.
The result is a more detailed and more richly articulated foundation for
handling NL within the OS framework, which will hopefully provide a
useful basis for further work in this area.
Finally, there is no reason why this enhanced OS framework should be
confined to the description ofNL as used within the context of organisations.
Rather, it offers itself as the basis of a more general-purpose approach to the
description of language as a semiotic system, and on such grounds it is all
the more worthy of attention.
REFERENCES
Connolly, 1.H., 1991. Constituent Order in Functional Grammar. Berlin: Foris.
Connolly, J.H., Edmonds, E.A., Guzy, 1.1., Johnson, S.R. and Woodcock, A., 1986.
Automatic speech recognition based on spectrogram reading. International Journal of
Man-Machine Studies, 24, 611-21.
Dik, S.C., ed. Hengeveld, K., 1997. The Theory of Functional Grammar, Part 1: The
Structure of the Clause. 2nd ed. Berlin: Mouton de Gruyter.
Halliday, M.A.K., 1966. Lexis as a linguistic level. In Bazell, C.E., Catford, J.C., Halliday,
M.A.K. and Robins, R.H., eds., In Memory of J.R. Firth. London: Longman, 148-162.
Liu, K., 2000. Semiotics in Information Systems Engineering. Cambridge: Cambridge
University Press.
Uu, K., Clarke, R., Stamper, R. and Anderson, P., eds., 2000, Information, Organisation and
Technology: Studies in Organisational Semiotics 1. Dordrecht: Kluwer.
Nauta, D., 1972. The Meaning ofInformation. The Hague: Mouton.
Stamper, R., 1996. Signs, information, norms and systems. In Holmqvist, 8., Andersen,
P.B., Klein, H. and Posner, R., eds., Signs of Work: Semiosis and Information Processing
in Organisations. Berlin: de Gruyter, 349-397.
Stamper, R., 1997. Organisational Semiotics. In Mingers, 1. and Stowell, F., eds.,
Information Systems: an Emerging Discipline? London: McGraw-Hili, 267-283.
Chapter 2
Means of Coordination
Peter B. Andersen, Peter H. Carstensen and Morten Nielsen
1. INTRODUCTION
The coordination of complex cooperative work is an intricate matter that

can impose a sever workload on the cooperating actors-so much so, in fact,
that the magnitude of coordination work can become a general obstruction to
work effectiveness, flexibility, or even safety. Therefore, systems for
computer supported cooperative work (CSCW) are often engaged as a means
of reducing the workload of the coordination activities. In order to arrive at a
design that reduces the magnitude of coordination work, some of the
coordination formerly performed manually by the cooperative actors will be
made part of the CSCW system; that is, certain coordination functions will
be supported or even fully automated by the computer system. Workflow
systems could, for example, be considered a system that reduces the
coordination workload by implementing a pre-specified protocol for the
routing of information, and, by providing structured forms, telling the user
what information he is expected to enter, etc. In order to establish a basis for
designing such systems we need a coherent conceptual understanding of the
coordination work conducted.
The CSCW literature is replete with on studies illustrating the rich and
seamless nature of coordination performed by co-located actors (Harper et
aI., 1991; Heath et aI., 1993). All these studies show that the coordination of
cooperative work can sometimes be achieved effectively and effortlessly
24 Peter B. Andersen, Peter H Carstensen and Morten Nielsen
simply by means of our everyday modes of interaction. Other studies have

shown however, that oral interaction becomes increasingly labor intensive
and sometimes even inadequate as a means of coordination in high
complexity cooperative work (Carstensen and Sorensen, 1996; Schmidt and
Simone, 1996). The latter studies highlight the need for artifacts that support
the actors in coordinating the complexly interdependent cooperative tasks
which, when carried out by means of oral interaction, tend to generate a very
high coordination workload. When designing CSCW systems with the
objective of reducing the magnitude of coordination work, it will therefore
often be necessary to port coordination work from the modality of oral
interaction to artifacts-this transformation is neither trivial nor well
understood.
This paper discusses how oral and artifact-based interaction among
collaborating actors is organized in order to reduce the coordination
workload. We relate and compare means for simplifying the interaction
applied for oral interaction to the structures and usage of artifact-based
coordination mechanisms. From doing this we can identify several
similarities that seem to indicate that it could be useful to approach both oral
and artifact-based coordination interaction in a systematic and theory-
supported way when aiming at understanding what goes on. Furthermore, we
argue that it could be relevant and useful to aim at a general set of concepts
and approaches for understanding coordination activities.
The process of moving coordination work from oral interaction to
computer-based artifacts fosters complicated issues as to the relation
between the two modalities of communication: How do cooperating actors
coordinate by means of oral interaction and artifacts respectively? And how
can we conceive of the transformation needed in order to arrive at a working
system that does in fact reduce the coordination workload? The present
paper seeks to contribute to the understanding of these issues by addressing
the similarities and differences between dedicated coordinative constructs
realized in artifacts and oral interaction, respectively.
We will start out by illustrating and analyzing a few fragments of oral
coordination as performed in maritime operations on board the world's
largest container carriers the MIS Sally Ma:rsk. We follow the work of the
crew during passages of manual steering where the oral interaction between
the actors serves to coordinate a host of complexly interdependent, time-
critical, and highly cooperative tasks.
After this we dig into the organization and structuring of the testing and
debugging of complex software. Here the coordination of the work of
handling and correcting bugs is supported by a form and a set of associated
procedures for the use of the forms. This is related to the conceptual
framework of Coordination Mechanisms (Carstensen, 1996; Schmidt and
Means of Coordination 25
Simone, 1996) founded on studies of coordination artifacts. The framework

provides insight into characteristics and qualities of artifacts designed
specifically to reduce the workload of coordinating complex cooperative
work.
In the last part of the paper we compare two types of coordination
methods.
It is our hope and intention that the analyses of oral and artifact-
supported coordination and the comparison between different media of
coordination-in a wider perspective-will provide input for establishing a
useful analytical approach to the analysis and modeling of coordination.
Before presenting the two field studies let us briefly introduce our basic
understanding of cooperation and coordination: The field of CSCW has
developed several different understandings of cooperative work and
coordination. In the following we will adhere to the notion that cooperative
work involves a number of interdependent actors.
Cooperative work is constituted by the interdependence of multiple
actors who, in their individual activities, in changing the state of their
individual field of work, also change the state of the field of work of others
and who thus interact through changing the state of a common field of work
(Schmidt and Simone, 1996, p. 158).
The concept of interdependence is at the heart of our understanding of
cooperative work and coordination. Interdependence sets the limits for what
we understand by the term cooperative work, and it shapes a basic
explanation for why coordination is an inescapable aspect of cooperative
work.
A group of people-a work arrangement-performing a set of tasks is
only to be considered a cooperative work arrangement if the tasks performed
are interdependent. Thus, the distinction between cooperative and individual
actors cannot be made by focusing on the actors themselves but on the work
they perform, cf., work related approaches like Strauss (1985), Carstensen,
(1996), Schmidt and Simone (1996).
It is the structure of the work-the interdependencies between tasks-
that binds the actors in a cooperative work arrangement together. The
interdependent tasks and the world of objects and processes, within which
they are performed, are referred to as the common field of work in an
attempt to underline the difference between isolated work phenomena and
work tasks that have a bearing on actors, and beyond the field of work of the
individual actor.
2. MARITIME OPERATIONS
The maritime operations with which we will be concerned in the

following were observed during the most recent in a series of field studies
concerned with time-critical cooperation and coordination as performed by
the crew on board some of the world's largest container carriers. We have
spent a total of four months on board three container carriers calling on 35
harbors on four continents.
The field work has generated rich data comprising a large number of
video recordings, interviews, and technical documentation on the vessels in
general and the instruments on the bridge in particular.
The data presented below stem from a European roundtrip performed on
board the carrier MIS Sally Mrersk during spring 1999. The fragments of
communication that will be analyzed in the paper have been transcribed from
video recordings. The overall objective of the analysis has been to establish
an understanding of the interdependencies between the cooperative actors,
the need for coordination, and the nature of the actual coordination
performed via oral interaction between the cooperative actors. In the
following we shall focus in particular on the communicative structures
engaged as a means of reducing the complexity of coordinating the complex
task interdependencies of maritime operations.
2.1 MIS Sally Mrersk
The MIS Sally Mrersk is the largest container carrier ever built: She is
347 meters long, 42 meters wide, weighs about 85000 tons when fully
loaded with a load counting some 7500 containers.
Figure 1. Sally Ma:rsk approaching the harbor of Algeciras

When traveling fully loaded and at maximum speed of 25 knots it takes

about 16 minutes and a distance of 6.1 kilometers to bring the vessel to a
complete stop. Even in ballast condition it will take 9 minutes and 3.2
kilometers to stop.
Basically, the movements of a sea-going ship are determined by the
interaction between controllable and uncontrollable physical forces. From
this perspective, navigation is concerned with applying the controllable
forces on the uncontrollable forces in a way that results in the desired
movement of the ship; the objective being to transport the ship and its cargo
between ports in a safe and efficient way.
Control able Uncontrolable

forces forces
.. ..
• Engine • Wind
• Rudder • Current
• Thruster • Swell
* Anchor • Sea
• Tug * Hydrodynamic
• Hawser effects
Figure 2. Controllable and uncontrollable forces determining the movements of the vessel
Ships the size and like of Sally Mrersk have quite limited
maneuverability. During the voyage it will often be the case that the
uncontrollable forces work strongly against the desired navigation, and at
times they might even be the most powerful. But even leaving aside the
influence of forces like wind and current, the vessel still remains the victim
of massive inertia-any change to speed or course takes time to build, and
once the change is initiated it is hard to arrest or reverse.
2.2 Overall bridge layout
The bridge of the Sally Mrersk is fully housed by a glass and steel
construction sheltering the delicate instrumentation and the actors from the
elements. The center of the bridge is the main working area for the
navigating crew; the bulk of instruments is located here. Most of the time,
operations are performed from the center bridge, except when in and around
the harbor basin where the vessel will be controlled via the instruments in
the port- or starboard bridge wing.
The bridge is fitted with four sets of manual rudder controls positioned
on the central console, helm stand, and the starboard- and port consoles,
respectively. Three engine telegraphs are available, one on the central

console, and one in each bridge wing.
Helmsman Radar Radar Voyage

stand Management
System
Ps. Central Console Stb.

Controles (engine telegraph) controles
Figure 3. The overall layout of the bridge
Figure 4. The central console of the bridge
During manual steering the maneuvering is manipulated hands-on by

either the master (captain) or the helmsman . The master can choose to
control both speed and course or he can engage a helmsman who will work
the rudder from the helm stand.
2.3 Cooperative work and coordination in maritime

operations
As mentioned in the introduction the concept of interdependence is at the

heart of our understanding of cooperative work and coordination. It is the
structure of the work-the interdependencies between tasks-that binds the
actors in a cooperative work arrangement together.
Means o/Coordination 29
Maritime operations involve many types of task interdependencies.

Consider, for example, the general nature of the work-interdependency
between the master of the Sally Mrersk and the pilot who assists operations
when entering and leaving harbor areas.
The fragments presented below may, at first glance, give the impression
that the pilot is in command of the vessel as it is he who issues the
commands. The fact is, however, that the master is in supreme command of
the vessel at all times-the pilot is granted the mandate of command but it
can be reclaimed by the master at any time and without further ado.
The division of labor between the master and pilot is implemented in
order to achieve the combination of two distinct specialties. The master
knows his vessel inside out, while the pilot is the expert on the local waters.
The pilot's job is to guide vessels safely in and out of the Rotterdam harbor,
he does so with several vessels a day, and is backed up by a network of other
pilots and the Vessel Traffic Management Service (VTMS) who are
responsible for the overall logistics of in- and outbound traffic. The pilot
carries a printed traffic plan and gets updated on changes via radio contact to
the VTMS; the pilot knows who is coming in and who is going out, at what
time, and what route the specific vessels are going to take. If direct
communication between two or more vessels should be needed, the pilot on
board the Sally Mrersk can contact his colleagues on the others vessel via
radio to make arrangements beyond the general traffic plan. Just as
important as his logistic knowledge is his experience in sailing the local
waters of the harbor area. The harbor is narrow, water depth is limited,
currents may run in unpredictable ways, and wind may change drastically
between locations in the harbor-the pilot knows about these matters and is
on board the Sally Mrersk in order to utilize his expertise for the safe
handling of the vessel.
The pilot's weak spot is that he does not possess in-depth knowledge and
familiarity with the individual vessel. The pilot works on many different
vessels rather than one, while the master operates this one vessel across
oceans and in and out of many different harbors. Safe and efficient
operations during the approach to Europort involves the expertise of both the
master and the pilot; the basic interdependence is a product of their work
specialization and the fact that the pilot is giving the orders while the master
carries ultimate responsibility for operations.
3. COMMUNICATION AS COORDINATION
Maritime operations are characterized by significant shifts in the need for

coordination. While operating in open waters the vessel is steered by one
officer who will perform most tasks in solitude; he is alone on the bridge
most of the time, and ship-to-ship and ship-to-shore communication is
infrequent. However, when the vessel approaches a harbor and enters coastal
waters, operations become highly cooperative. When operating in and
around the harbor the master (captain), chief officer, helmsman, and pilot are
working on the bridge; two groups of three and four actors--commanded by
the first- and second officer, respectively-are positioned on deck; the pilot
station, vessel traffic management service, and dockers are located ashore
and communicate via VHF; tugboats and other vessels operate in the
immediate surroundings of the Sally Mrersk and will at times become part of
the cooperative work arrangement involved in maintaining safe operations.
Cooperation grows in line with rises in work complexity. When the
vessel moves from open waters through coastal waters on to the waters in
and around the harbor, work constraints become increasingly tight.
Basically, maneuvers have to be performed with an increasing degree of
precision, while there is less time to make the navigational decisions.
The tighter work constraints are reflected in the ways the actors
communicate and coordinate. In and around the harbor area-where work
complexity peeks--communication is about work only and proceeds
according to well established patterns. It is this type of coordination-the
oral coordination of complex cooperative maritime operations- that we
shell be concerned with in the following. The theoretical basis for the
analysis will be drawn from empirical and theoretical work on work
communication (PB Andersen, 1997). We shall characterize instances of
work related communication by means of the key concepts: focus,
background, and protocol.
3.1 Focus and background
Consider the following situation. The Sally Mrersk is inbound for

Rotterdam. There are four actors on the bridge, master, chief officer, pilot
and helmsman. The auto pilot has been discharged and steering is performed
manually. All are silent, then the pilot speaks: starboard twenty. The
helmsman speaks: starboard twenty. And all are silent again.
1 Pilot Starboard twenty

2 Master Starboard twenty
Fragment 1: Rudder command

In any communicative situation, some features are taken for granted,

whereas others are the topic of discussion. We shall say that the former
constitute the background, the latter the focus of communication 1.
Background information is either not expressed at all, referred to via
unstressed pronouns or definite nouns that recur identically during the
conversation, or referred to via certain backgrounding constructions, such as
adnominal adjectives and clauses.
In the communication between the pilot and the helmsman, the
background information is completely left out. The pilot does not state who
is the receiver of his command (the helmsman) nor which action is to be
done (turning the wheel); he does, in fact, not state that it is a command at
all.
Particular combinations of backgrounded and focused information types
define specific communicative functions. The communicative functions can
be defined in a very precise manner by considering a typical sentential
schema used to interpret actual utterance~.
Subject Tense Verb Object Manner Time, place

Aspect
Modality
Someone do acting on Something in some way sometime
Schema A. The basic sentential schema.
1 Focus and background are well-known concepts in linguistics. They describe the
information structure of the sentence, cf. the textual metafunctions in Halliday 1994.
2 The approach is mainly inspired by Halliday 1994. According to Halliday, utterances must
simultaneously fulfill three main types of functions: ideational (how should we structure
the topic?), inter-personal (how do we interact?), and textual (how do we make the text
cohere?), cf. note 1. The schema tries to capture the combination of functional features
that are relevant in the data. Ideational functions: the schema only covers Doings, i.e.
material processes implying a change of state. Halliday's five other process types are not
covered. The structure, Subject (Actor) + Verb + Object (Goal), derives from the process
type of Doings. The adjuncts Manner, Time and Place are added since they turned out to
be important in the data. Interpersonal functions: A slot for tense, aspect and modality is
included since we are concerned with co-operation, i.e. regulation of interpersonal
relations. Textual functions: the distinction between background and focus is a textual
function. It is included in order to describe the reduction of complexity so important in
complex work settings. Thus, the schema is an assemblage of features that are expected to
be important in cooperative, physical work of the type we are dealing with. The schema
may not be able to extract important features of other types of work, e.g. work mainly
consisting of communicating - "sayings" in Halliday's terms.
A sentential schema consists of a set of slots following one another in a

more or less fixed sequence. Each slot can be filled with a particular kind of
linguistic material, e.g. nouns, verbs or adverbials. Such sets have been
called paradigms since Saussure's days. In the following diagrams we shall
shorten "Tense, aspect, modality" to TAM.
Below a set of communicative functions are defined by different
combinations of background and focus (italic) paradigms (the functions are
defined based on data from the Swedish Postal Giro, cf. Holmqvist (1989)
Andersen (1997), pp. 379 ff.):
I Someone I should begin I acting on I Something I in some way I sometime
Schema 1. Work distribution: who should begin doing it when?
I Someone I acting on I Something I in some way I now
Schema 2. Supervision: is she doing it?
I Someone I acted on I Something I in some way I now
Schema 3. Worker Reporting: has she done it?
I Someone I has I acted on I Something I in some way I now
Schema 4. Object Reporting: has it been done to it?
I should I act on I Something I in some way I now
Schema 5. Work coordination: what should I do now?
I should I act on I Something I in some way I sometime
Schema 6. Instruction: how should she do it?
Someone has acted on Something in the sometime

correct way
Schema 7. Control check: has it been done to it in the correct way.
If we consider the rudder command communicated between the pilot and

the helmsman in terms of the basic sentential schema, only the manner slot is
focused (Schema 8) the rest of the slots are backgrounded. Thus it belong to
the instruction type.
Subject TAM Verb Object Manner Time, place

Helmsman should tum the wheel Starboard Now, at the
twenty helm stand
Schema 8: Instruction. Helms man should tum the wheel in some manner now at the
helmstand
The purpose of using background and focus paradigms is to reduce

complexity. If the utterance is to specify an action that can be performed,
then all paradigms must contain exactly one member that connects to work
operations. One cannot make the ship sail "some" course. The course must
be "001°,002°,003°, etc.". So the purpose of the conversations is to fix all
focus paradigms to exactly one of these. But if we can choose between, say
10 items in each slot, there are 106 possibilities to choose from, and we will
never get around to do anything before the ship is grounded.
However, if we do not have to bother with the subject (the helmsman
does it), the time (it is now), the action (tum), the object (the wheel),
complexity is reduced to the manner slot with a few degrees of freedom (the
wheel can be moved 30 degrees to each side). Formally speaking, the
technique of backgrounding reduces complexity exponentially.
Bridge work builds on a well established and predefined division of
labor. One basic rule is that there is only one actor-the actor in command-
who can issue commands. The other actors can suggest actions but they do
not become commands before spoken by the actor in command. This
division of labor can be defined as a backgrounding of the actor in command
who is only focused when the watch shifts: "Will you take over now?"
When approaching a harbor, the command of the vessel is first passed on
from the on-duty officer to the master (when initiating cooperative
operations), and then from the master to the pilot during the final stages of
the approach. The pilot receives command of the vessel shortly after
embarkation and keeps it until the vessel is made fast at the quay. During
this period the vessel is most often steered manually via one of four steering
consoles. Only one console can be active at a time. The master has hands-on
control of three consoles, while the helmsman operates the last console, the
helm stand. If the helm stand is in operation and the pilot has command of
the vessel, then the helmsman knows that he is the presumed addressee of a
rudder command issued by the pilot. Conversely, if steering is performed
from one of the three other steering consoles, the addressee is the master,
etc.
Typically very few paradigms are focused. The most frequent are the
manner (cf. Schema 8), the time and the aspect/modality paradigm. Many
maritime conversations are about the detailed parameters for handling
instruments, the time for doing an action, and whether an action is imminent,
executing or ended (aspect). Fragment 2 shows an example with focus on
time
I C officer ... half an hour before

2 2. officer Half an hour before; that is a Quarter to
3 C officer Yes, it will have to be, if the pilot is not boarding until a Quarter past
4 2. officer I guess so
Fragment 2. Focus on time.
I C. officer Yes, ok, thank you N - you can go fore and aft now [radio]
2 2. officer We are readv aft [radio]
3 Master You are ready aft [radio]
4 1. officer We are ready on the Forecastle [fore mooring deck] [radio]
5 Master You are ready on the Forecastle fradio]
6 Master You can single up to one and one [radio to the officers on the fore and
aft mooring decks.
7 2. officer We single up to one and one [radio]
8 I. officer One and one fast fore [radio1
9 Master And that was one and one fore [radio]
10 Master All singled up for and aft Sir
II Pilot Okay
12 1. officer ( ... ) fore hawser I!:one fradiol
13 Master Fore hawser I!:one fradio1
14 2. officer And the aft spring is I!:one fradiol
15 Master Aft SI>I"ing gone [radio1
16 Master Yeah, I guess we can let him let him take the stem line ( ... ), hanging in
the forward spring
17 Pilot Yeah, okay
18 Master Just let the aft hawser go [radio1
19 2. officer Let go the hawser aft [radio]
20 Master You can wheel that in at the same time can't you [radiol
21 2. officer yes [radio}
22 C. officer He will have it wheeled in before, before they [the thrusters] get to it I
guess
23 Master Yes, just about so, it's not going very fast
Fragment 3. Orchestrating departure.

Fragment 3 below contains utterances concerned with removing the

hawsers and lines during the first stage of a departure from harbor. The
fragment involves the following actors on the bridge: master, chief officer,
and pilot. And the first- and second officer on the fore- and aft- mooring
decks.
The words in boldface signify modality (can), tense and time (will have,
before), or aspect (imminent: ready, let; executing: are wheeling; ended: the
present perfect, e.g. all gone, (made) fast, singled up, etc.). Note also time
indications that follow a quasi ready-steady-go pattern also found in some
sports.
Schema 9 describes the sentential schema underlying Fragment 3. It is a
combination of reporting and supervision and may be termed Orchestrating.
Subiect TAM Verb Object Manner Time, place

I. and 2. Preparing, Letting go, Springs and Fore and aft
Officer + starting, pulling in. hawsers now
ship executing,
assistents ending.
Schema 9. Orchestrating: will he do it, is he doing it, or has he done it?
In Schema 9, the hierarchy and the phase of the voyage has filled out all
slots except the modality and time slots. For example, it is the right of the 1st
officer to be located afore and not aft!
Such communication types, with many backgrounded paradigms, are
normal in work communication (Falzon, 1983; Falzon, 1984) but it is always
accompanied by another type, where suddenly the fixed background
paradigms "thaw out" and become focus paradigms. What was previously
taken for granted suddenly becomes the subject of discussion. The ensuing
utterances increase in complexity since they contain more focus paradigms,
and therefore open more choices (cf. the distinction between language in and
about a situation (Halliday, 1978) and the participant versus spectator
perspective (Holmqvist, 1989: 80).
Shifting from participant to spectator perspective is difficult but highly
important, since it must happen when the assumptions hidden in the
background paradigms are not valid any longer. The captain on the ship has
this ability, as the following example shows.
Fragment 4 is from an approach to the Harbor of Rotterdam where the
Sally Mrersk has been forced to perform a 360 0 tum in the deep-draft
fairway some five miles off the harbor entrance. The maneuver is critical
because no vessels can pass her while she is sideways in the canal. Traffic
conditions are cleared prior to the maneuver. Yet, just after having initiated
the turn, the Sally Mrersk is contacted on the VHF by the outbound Seal and
Atlantic that will meet with Sally Mrersk in the narrow canal.
1 S.Atlantic Sally Mrersk, the Sealand Atlantic [VHF]

2 Pilot Sally Mrersk [VHF]
3 S.Atlantic Yeah, good afternoon Captain, are you turning to port now, are you,
over? [VHF]
4 Pilot Yes, I'm turning slowly to port, yes [VHF]
5 S.Atlantic Okay, we, we are, we will be steering our course of about two nine
zero, and we will stay to the north of you, if that is agreeable with you
[VHF]
6 Pilot yeah fine, I will be following the deep draft route outside [VHF]
7 S.Atlantic Yeah, and can you give us a red to red passing, please, port to port
[VHF]
8 Pilot port to port, yeah fine okay [VHF, talking to the Sealand Atlantic]
Fragment 4. Focus on parameters.
The assumed background action is "passing head-on" and the focus

paradigm contains two members, red to red and red to green, from which
they have to choose. Sealand Atlantic chooses red to red because this is the
normal choice. However, since Sally is turning port, the action will not be
"passing head-on" but "overtaking" where red to red is a wrong choice.
Sealand Atlantic knows that Sally is turning port, but has not realized the
consequences (Fragment 5). The pilot does not realize the error either, but
the captain does. He begins to focus the modality paradigm (how can we)
and the previously backgrounded action slot (he will overtake us probably)
and the Pilot quickly realizes his error (not possible) - although he does not
admit to it!
Fragment 5 is a good example of how important it is that speakers are
able to focus what was previously a background paradigm. On the one hand,
it is necessary to reduce the complexity by backgrounding information, but it
is risky too if the backgrounded information is not valid.
The examples given above can be summarized into the following
observations regarding oral communication and coordination:
In oral communication, information is distributed in background and
focus paradigms. This enables the actors to reduce complexity of their
interaction.
Coordination communication consists in closing the focus paradigms so
that one executable member remains.
Communicative functions can be defined by combinations of background
and focus paradigms.
Oral communication allows smooth changes between focus and

background paradigms, enabling speakers to question tacit assumptions.
1 S.Atlantic Standing by zero two one six[VHF]

2 Pilot okay [VHF]
3 Master who was that?
4 Pilot the outgoing ship, Sealand Atlantic, she want to pass port to port
7 Master How, how can we do that? [pass the Sealand Atlantic port to port]
8 C. officer (... )
9 1. officer Yes
10 Pilot We are steering around slowly, slowly
11 Pause
12 1. officer I presume he means that he will ...
13 Master Yeah, but how can he ... (. .. ), he will overtake us probably ....
14 Pilot Yeah, I don't know
15 Master ... yeah, no, so urn, port to port
16 Pilot Sealand Atlantic, the Sally Mrersk [VHF]
17 S.Atlantic Sally Mrersk, Seal and Atlantic [VHF]
18 Pilot Urn, you want to pass us on our starboard side, on the north side?
[VHF]
19 S.Atlantic Roger, III like to .. .I, I heard you were turning to your port to go back
in, I ( ... ) in with the drencher, otherwise we are going red to red, over
[VHF]
20 Pilot I think red to green with us [VHF]
23 PilotSt Sealand Atlantic ,Pilot Maas [VHF]
24 S.Atlantic Pilot Maas [VHF]
25 PilotSt The, the Sally Mrersk is turning to port, so you can proceed her north
of her, over [VHF]
26 S.Atlantic she is gonna turn to port, okay thank you, thank you, Sally Mrersk
[VHF] [notice that the Sealand Atlantic thinks they are talking to the
Sally Mrersk-when they are in fact talking to the Pilot Station Maas
Approach].
27 Pilot yeah ( ... ) port to port, red to red not possible
28 Master that is not possible
29 Pilot his red side to our green side
Fragment 5. Shift offocus from parameters to modality.
3.2 Protocols
As is well-known, communication is realized as a sequence of utterances

governed by certain tum-taking rules of the form
1. If someone is saying something to me, I should say something to

someone.
An example from the maritime domain is:
2. If! receive information (e.g. via VHF or intercom), I should repeat or
summarize it in order to prevent misunderstandings.
However, other events than communicative ones are governed by
conventions. In (3) it is a physical action that triggers speech, in (4, 5) a
perception:
3. If! have changed the controls on the bridge, I should tell it to my
colleagues.
4. If I can see something relevant which my colleagues cannot see, I should
inform them of it.
5. If I notice someone making an error, I should inform him of it.
Ronald Stamper and his collaborators (Chong & Liu 2000) claim that
such rules - or norms, as they call it - are a vital ingredient in the proper
functioning of work organizations. They set up two main types of norms,
behavioral norms and intrinsic norms, comprising evaluative, perceptual, and
cognitive norms. (2)-(5) are examples of behavioral norms that have the
form:
6. If <condition> then <agent> is <obligedlpermittedlforbidden> to perform
<actionlspeech act>
An intrisic norm has the following form:
7. If <condition> then <agent> adopts <an attitude> towards <some
consequences or proposition>
Example:
8. If someone is hindering me to do something then I expect him to notify
me when the coast is clear.
Norms in our data are heterogeneous since general and specific norms are
mixed; they are semi-conscious since some of them are written down and
others are verbalized when they are broken; and they are not designed as a
system, but have evolved through history.
The existence of norms can be verified in four ways: they can be
abstracted from regularities in the data; they are verbalized when they are
violated; they are taught explicitly during training; and some norms occur in
the ship-owners manuals.
Norm (2) can be extracted from the data in Fragment 3 and norm (4)
becomes explicit in Fragment 6:
I Master And call out when you are on the new course, K, right ( ... ), when it's
there you say one one five - then we know it's there
2 Helmsman I did say so last time
3 Master Well, Ijust did not hear it
4 Helmsman (.. )
5 Master okay (... ) that's fine - keep up the good work
Fragment 6: Violating rule (4).
Norm (8) surfaced when a ship was delayed in the berth which Sally
Maersk was scheduled for; the ship left the berth without notifying Sally,
which caused some grumbling aboard. (3) could be observed indirectly
several times when the captain asked to pilot to keep his hands off the
controls: this reason for this is that pilots forget to say what they are doing
and therefore violate norm (3).
Finally, (5) occurs in the ship-owner's manual as the formulation:
"subordinates MUST SPEAK UP when orders in their considered opinion
are contrary to rules and regulation or will obviously give results contrary to
the superior's intent". Thus, norms are real and can be empirically verified.
Norms work concurrently as well as sequentially. The former is the case
where a single utterance or act is the result of more than one norm; the latter
occurs where a norm adds a new action or utterance after the previous one.
For example when the helmsman receives a course-order, (2) is responsible
for his repetition of the order, (4) causes him to report when the new course
has been reached, and (2) makes the master repeat the report. Sequential
application of simple norms can generate long complex sequences.
Many of the maritime norms are rooted in the fact that all actors are
obliged to take active part in monitoring the state of affairs in the field of
work, the maneuvering of the ship. In this respect, the maritime field of work
is common to all the cooperative actors-if a state change is performed by
one actor, it is imperative that the other actors are informed of the change.
During work norms are only verbalized when they are endangered and
therefore need maintenance. We have already seen this in cases of violation,
but less dramatic examples also occur, e.g. when colleagues remind one
another of their duties. In this case the crew often communicate about how to
communicate: the norms are self-referential (Fragment 6,7 and 8).
Fragment 7. Self-referential communication.

1 C officer Have you talked to the pilot [radio]

2 2. officer No, I'm about to call him [radio]
3 C officer Then ask him what side [of the vessel] he wants the pilot-ladder on
outside [the ladder that the pilot uses to get to the pilot boat when he
disembarks]
4 2. officer Yes, okay [radio]
Fragment 8. Self-referential communication
As has appeared from the preceding, norms are modal, i.e. they contain
words like is forbidden, allowed, possible, impossible, probable, expected,
etc. In some areas, norms are conventionalized: for example, the shape of a
traffic sign indicates whether it prohibits, demands, or allows an action, or
just gives information.
The norms for oral interaction have a number of characteristics:
They are semi-conscious, i.e., normally not conscious, but can be
verbalized and discussed in case of violations. Formulations can also be
found in maritime education and in ship-owner's manuals.
Norms do not form a system, but are a set of concurrently working,
heterogeneous, and relatively autonomous guidelines that have evolved
through time.
Protocols can communicate about themselves (are self-referential).
A special subset of norms are those concerned with regulating
cooperation between two or more persons. We shall use the term protocol to
denote a coherent set of norms of this type. All the preceding examples of
norms belong to a protocol.
4. COORDINATING ARTIFACTS
Let us now change the scene completely: We are at Foss Electric, a

Danish manufacturing company developing, producing, and marketing
equipment for, amongst others, measuring the compositional quality of milk,
a highly specialized field. Development and production is localized in
Denmark, sales, service and distributors are spread all over the world. The
Foss Electric Corporation employs approximately 700 people.
Foss Electric has. implemented concurrent engineering cf. e.g., Helander
(1992) yielding integration between manufacturing functions throughout the
development process and the organization is very much structured in terms
of projects including specialists with competence in fine mechanics,
chemistry, hardware and software design. The study to be discussed here
concentrated on the software development in one large project: The S4000
project. It was a highly complex design: It included facilities earlier provided

by several instruments, measurements of new parameters in the milk were
included, speed was to be improved significantly, and it was the first product
with an Intel-based 486 PC build-in.
4.1 Software testing and correction work
The software complex contained more than 200,000 lines of C-code and
was organized in approximately 25 modules distributed in 15 different
application. The S-4000 project involved more than 50 different people and
lasted approximately two-and-a-half years. During the last 18 months the
software design group had a stable size of approximately 8-10 designers,
each having rather clear roles related to the design and implementation of
one or more specified modules.
The most import roles related to the testing and correction activities
included: (1) Software designers responsible for designing, implementing,
maintaining, and correcting bugs·in one or more of the software modules; (2)
A group of three software designers called the Spec-team responsible for
diagnosing reported bugs and deciding how to handle each of the bugs; (3) A
Platform master responsible for managing and coordinating all the activities
involved in integrating the outcome of one working period (called a
"platform period"). He was, among other things, responsible for verifying
the corrections of the software made by the designers, i.e., control that the
reported bugs had been dealt with; (4) A Project plan manager responsible
for maintaining a project plan spreadsheet; (5) Testers testing the software
embedded in the S4000 instrument; And (6) the central bugs file manager
organizing and maintaining the central bug file, a ring binder containing
copies of all reported bugs and organized according to their status. It is
interesting to notice, that there was no software group manager during this
one and a half year period.
Early in the S4000-project the software designers realized problems in
coordinating, controlling, monitoring, and handling the testing activities.
They invented and used a standardized bug form that all testers had to fill in
whenever they identified an error (a bug). To prescribe the use of the forms,
a structured ring binder (being used as a central file) and a set of procedures
and conventions for the use of the form were established. Some of the
procedures were written down as organizational procedures, others were just
conventions developed and refined during the project. It was the recognition
of the mentioned problems that also led to the establishment of the roles
mentioned above. These were defined in order to establish the basis for
running the procedures.
The aim of this section is to illustrate how the use of the form and
concomitant work procedures were used in order to cope with the
complexity of coordinating certain activities in relation to reporting,
diagnosing, and correcting software bugs.
The development and correction work was organized in phases called
"platform periods". A platform period was typically 3-6 weeks work
followed by one week of integration. All the work and the plans were
structured in relation to these periods. For each period a designer was
appointed Platform Master responsible for collecting all information on
updates and changes made to the software, and for ensuring that software
was tested and corrected properly.
When a bug was identified, the tester filled in a form and sent it to the
spec-team. The spec-team diagnosed the problem and decided which
developer should fix the problem. The responsible designer was notified (by
receiving a bug form), and estimated the correction time needed. When the
problem was dealt with, the designer notified the Platform Master who could
then verify the corrections.
At any state, the binder contained a copy of the form in its current status.
The binder had seven entries reflecting the status of a specific bug: (1) Non-
corrected catastrophic bugs (copies); (2) non-corrected essential bugs
(copies); (3) non-corrected cosmetic bugs (copies); (4) postponed bugs
(originals), (5) rejected bugs (originals); (6) corrected bugs not yet verified
(copies); and (7) corrected bugs (originals). For each of the seven categories
the forms were filed chronologically. The entries played a central role in
stipulating the coordination by providing all involved designers and testers
access to the state of affairs in the testing.
Initials:
Date: (I)
Instrument:
I Report no: (2) Th~ !!~tors fill {or adg information) in:
The testers: (t), (2), (3), and (4)
Description: (3) The Spec-team: (3), (4), (5), and ( 7)
The designers: (6) and (8)
The Qrocedure for handling bugs:

Classification: (4) • A tester registers and classifies a bug
1) Catastrophic 2) Essential 3) Cosmetic (field 1,2,3, and 4)
• The tester sends the form to the spec-team
Involved modules: (5) • The spec-team diagnose and cia ssifies the bug
Responsible designer: Estimated time: (field 3, 4 and 7)
• The spec-team identifies the
Date of change: Time spent: Tested date{6)
o Periodic error - presumed corrected
responsible designer
(field 5)
• The spec-team estimates the co rrection time
Accepted by: Date: (7) (field 5)
To be: • The spec-team incorporates the correction
1) Rejected 2) Postponed 3) Accepted work in the work plans
Software classification (1-5): ___ • The spec-team requests the desi gner to
Platform: correct the problem
• The designer corrects the hug and fills in
Description of corrections: additional correction infonnatio
(8)
(field 6 and 8)
• The designer sends the form to the central file
Modified applications: • The CFM sends the form to the PM and
insert copy in central fi Ie
• The PM verifies the correction
Modified files: • The PM returns the form to the central file
Figure 5. A translated version of the bug form and the procedure followed when using the
forms. CFM is central file manager and PM is platform master. The form is a sheet of A4
paper printed on both sides. The figure illustrates who fills in the information in the form.
The work plans were organized in a large spread-sheet containing

information on: which tasks are to be accomplished and a reference to a
detailed description of the task, the estimated amount of labor-time per
module for each task, the responsibility-relations between modules and
software designers, the relationships between the tasks and platform periods,
and the total planned work hours per platform period for each software
designer. The work plans were maintained by the Platform Master and a
member of the spec-team.
The usage of the forms and the procedures can also be viewed as
information flow between the involved roles and actors. The figure below
illustrates the flow between the six different roles involved in coordinating
the software testing and correction activities. In many situations the involved
actors undertook more than one role.
1 ~ \ ___
!
Central
file manage
~
~c-tev
4~~ 5
Figure 6. A visualization of the roles involved in the software testing of the S4000 project,
and the information flow between them.
The infonnation flow described in Fig. 6 concerns only the stipulated

(through organizational procedures) flow of the bug fonn mechanism. The
thick arrows (3, 4, 6, and 8) indicate, that the flow is often a bunch of fonns
sent. The thin arrows indicate, that the fonns typically are sent one at a time.
Other types of infonnation are frequently exchanged between the actors and
there might be situations were one of the actors choose not to follow the
stipulated infonnation flow.
4.2 Coordination of software testing and correction
Software testing and correction is a very complex and complicated task

requiring a lot of coordination. In order to avoid redundant work, the testers
need to be aware of each others work. The study showed, that it was very
difficult for the testers to get an overview of reported errors, their diagnoses,
correction status, etc. It was difficult for the actors involved in the S4000
project to detennine the state of affairs in the software testing at a glance,
and it was difficult for the testers and designers to communicate about the
software complex and its status at a given point in time.
The coordination activities within software testing and correction were
mainly based upon conceptualizations of structures in the field of work (e.g.,
the structure of the software complex) and structures reflecting the current
implementation of the cooperative work arrangement (e.g., the involved
actors, the working cycles, verification procedures, etc.). The
conceptualizations were, among other things, used to support the distributed
bug registration activities, support the planning activities, monitor progress
in correcting the software, monitor the state of affairs in general, and to
simplify the needed bug classification and diagnosing activities.
Furthermore, aggregations of detailed information on the state of affairs
(e.g., the total number of "not yet corrected category 2 bugs") was used to
support the coordination work, especially in order to simplify the required
monitoring activities. Several structures for classification and categorization
of bugs, corrections demands, and software modules were used. Concrete
information from the software testing and development was also used when
the activities were coordinated, e.g., the software code itself, the
documentation, or the content of the bug registrations were used to decide
the estimated correction time for a bug.
4.3 Coordination support provided by the bug report

mechanism
It can, of course, be questioned whether the bug form mechanism

genuinely has "eliminated" coordination complexity. The coordination to be
conducted is as complex as always, but it can appear simpler to the actors
through improving the representations of the work domain, by enforcing a
specific behavior of the actors involved, by "automating" certain activities of
the coordination, and by establishing a division of labor minimizing the need
for coordination. The claim here is, that the bug form mechanism supported
the coordination of software testing and correction by providing several of
these things:
- The bug form provided a standardized information structure by which all
bugs were described. By allowing information to be used for the
diagnosis to be included in the form in a standardized pre-structured
manner, the mechanism made it easier for the spec-team to find the
relevant information. The classification of the bugs made it easier for the
spec-team to deduce the testers' perception of the problem reported. The
form can thus be seen as improving the representation of the field of
work (the bugs) by establishing a "common standardized language". This
makes it easier for the actors to interact. "Standardized languages" are, of
course, problematic too. They constrain the actors, and it takes time for
the actors to become familiar with them, and they need to be maintained.
- The standardized format of the form, furthermore, supported the work of
reporting, both from tester to spec-team, from spec-team to designer, and
46 Peter B. Andersen, Peter H. Carstensen and Morten Nielsen
from designer to platform master. This is because it forces a specific

behavior on the actors. Through a specific surface representation the
form supported the actors: No effort was required for considering which
information to include.
The bug form mechanism also supported the coordination activities by
stipulating the work flow for handling the reporting, diagnosing, and
correcting process. Although it was not completely automated, the pre-
specified flow minimized the need for communication and interaction
among the actors when handing over the form (and thus the obligations)
from one actor to the next. The pre-specified flow (the embedded
protocol) afforded support through constraining the actor: He could just
apply the pre-specified routing without further considerations.
- Finally, the bug form mechanism was a central tool in the attempt to
establish a well-understood and well-defined division oflabor. By
establishing different roles, and very clearly defining their
responsibilities, the mechanism reduced the need for communication and
interaction among the actors. All actors in the reporting, diagnosing, and
correcting process knew exactly what their obligations were. When they
had dealt with their part of the treatment, they could just pass on the
form, and others would take care of the rest. Their need for coordination
was reduced.
Coordination activities, like monitoring the state of affairs, and
negotiation of classifications, allocations, etc., were attempted to be
supported by the bug form mechanism too. The establishment of the central
file (the binder) including all registered bugs and their current status made it
easier to get a coherent picture of the state of affairs. Although the testers
and designers found it difficult to achieve an overview from the content of
the binder, the benefit was that they only had to search in one place.
Furthermore, the standardized information structure in the forms, and the
standardized index of the binder, made it easier for the actors to find the
relevant information, for example the classification and status of a bug or the
number of not-yet-corrected category 2 bugs. Regarding negotiation of bug
classification or resource allocation the bug form mechanism, and its related
mechanisms, made classifications and resource allocations visible and
accessible to the actors. The classifications and allocations became easier to
discuss.
4.4 Artifact-based coordination
The example of the bug report illustrates that artifacts can be effective
means of reducing the complexity of coordinating complex cooperative
work. The issue of how artifacts can be instrumental in reducing
coordination complexity has been thoroughly addressed in many insightful

publications (Malone et aI., 1987; Fitzpatrick et aI., 1995). While inspired by
the above mentioned, the analysis of the bug form mechanism has
contributed to a conceptual framework called Coordination Mechanisms
(Carstensen, 1996; Schmidt and Simone, 1996).
The aim has been to develop a set of concepts for analyzing,
understanding, and modeling coordinative artifacts as observed in real world
work settings. The framework addresses both analytical and constructive
aspects of artifact-based coordination.
As briefly introduced in the introduction, our understanding-and the
conceptual framework of Coordination Mechanisms-is based on an
analytical distinction between the cooperative work arrangement and the
common field of work, and a distinction between work and articulation work
(Schmidt, 1994; Carstensen, 1996; Schmidt and Simone, 1996). The first
analytical distinction is important when identifying who are collaborating
(i.e., mutually interdependent) and with respect to what they are
collaborating. The latter distinction is needed in order to distinguish
coordination from the 'ordinary' work activities that will have to be
performed no matter if the work is performed cooperatively or not. The
concept of coordination mechanisms has been established in relation to these
analytical distinctions. A coordination mechanism is a dedicated
coordinative construct -the coordination mechanism is aimed specifically
at supporting the coordination needed in order to bind together the
individual, yet interdependent, tasks performed by the cooperative work
arrangement. Coordination mechanisms are objects having a physical form
and they achieve a reduction of coordination workload by stipulating and
mediating coordination. The coordination mechanism stipulates who is to do
what, where, and when, while also mediating the information needed by the
individual actors.
From the analysis of the bug form mechanism and the work on the
conceptual framework a number of relevant characteristics of artifact-based
coordination mechanisms have been identified:
An artifact-based coordination mechanism is essentially a protocol. It is a
set of explicit procedures and conventions that stipulate the coordination
of the distributed activities. The distributed activities are coordinated by
"executing the protocol".
The stipulations of the protocol are (partly) conveyed by a symbolic
artifact, i.e., they are persistent and publicly available in some kind of
'physical form'. Thus, a cognitive symbolic structure only existing in 'the
head of the actors' is not considered a coordination mechanism.
It is the symbolic artifact that mediates the coordination of the distributed

activities. The artifact is an intermediary between the collaborating
actors.
- An essential feature of an artifact-based coordination mechanism is that it
has a standardized format that reflects pertinent features of the protocol.
It thus provides affordances to--and impose constraints on-the required
coordination.
- The state of the protocol embedded in the artifact-based coordination
mechanism is distinct from the state of the field of work, i.e., it contains
conceptual representations of aspects of the field work and the work
arrangement. Changes to the state of the field of work are not
automatically reflected.
5. COORDINATION MECHANISMS AS REIFIED

ORAL COMMUNICATION
Ifwe compare the communicative functions described in Section 3 to the

functions realized in the bug-report from the previous section, we find most
of them again.
Parts of the bug-report can be analyzed as a medium for conducting a
sequence of coordinative communicative functions with different focus and
background paradigms. In addition, the focus paradigms in one function
becomes background in the following. As shown in Schema 10, the verb is
always backgrounded to correct; (1) focuses on the module identification,
(2) on modality (should/should not be corrected/rejected/accepted), time
(now/postponed) and manner (classification of bug: how should the
correction be done?), and (3) focuses on the subject (who is going to do it).
Finally, (4) is concerned with aspect (has it been done) and manner (has it
been done in the proper way).
1. Someone should correct which in some way sometime
module
2. Someone should/ correct the module in which when
shouldn't way
3.Who should correct the module in that way at that time
4. Someone has corrected the module in the at some time
proper
manner
Schema 10. Coordinative functions in the bug-report in terms of foreground / background

focus paradigms
If we compare the paradigms to the different combinations presented in

Section 3.1 we can see that (3) is Work Distribution (Schema 1), (2) is a
variant of Instruction (Schema 6) where modality is focused too (should/
shouldn't be done), and (4) is Control Check where tense (has it been done?)
and manner (here "correctly") are focused.
The first function (1) is a new one (compared to those introduced in
Section 3.1) where only the work-object is focused. Also the "seriousness-
classification" (catastrophic, essential, cosmetic) lack analogs in the data of
Section 3.1, although it resembles the Work Priority Ordering defined ilJ..
Andersen (1997: 375).
We can conclude that communicative functions found in oral
communication recur in coordination mechanisms. The bug-report collects a
set of communicative functions that are normally applied after each other
according to the oral protocol, and their sequence on the form to some
degree stipulates their temporal sequence. For example, the Control Check in
(4) will normally be performed after the Work Distribution in (3).
The bug report is thus a reified version of the oral coordination.
In our two examples, the main difference between oral and artifact based
coordination is that the present state of the field of work, its history and
possible futures are made persistent and thus publicly accessible. Whereas
the software developers at any time can inspect the forms to ascertain how
far work has progressed, what has already been done, and what might
probably happen in the future, the situation awareness of the maritime
officers must continually be reproduced and updated. Thus, the developers
seem to be better off than the officers and one may wonder why persistent
coordination mechanisms are not utilized more on the bridge: the only
persistent mechanism we could find on the bridge were the checklists used
before voyage start.
The reason is that there is another side of the coin. When coordination
means become persistent they are more difficult to change and adapt to
current conditions.
As stated in much of the work on coordination mechanisms (Carstensen,
1996; Schmidt and Simone, 1996), adaptability of coordination mechanisms
is a desideratum for coordination artifacts too. Therefore the protocol of a
coordination mechanism must be publicly available and malleable. Although
the bug-report form obviously reduced part of the work-load of handling the
testing and correction tasks, it was also clear from the study that situations
occur where the protocol is not applicable, and that the flow-structure (and
thereby the protocol) changed over time. The consequence-at least in the
case of computer-based coordination mechanisms-is that much of the
underlying protocol needs to be acquired and explicated in some form.
This particular facet of artifact-based coordination support has made the
concept of protocol controversial within the field of CSCW research-and
for good reasons. Schmidt and Simone discuss the notion of protocol from a
sociological point of view and state that
"study after study have demonstrated, unambiguously and beyond any
doubt, that the status of these formal organizational constructs in the actual
course of work is problematic in that these constructs are impoverished
idealizations when taken as representations of actually unfolding activities"
(Schmidt and Simone, 1996: 166).
One reason why persistent coordination mechanisms do not play any
important role in maritime navigation was in fact already given in Fig. 2: the
work consists in pitting the controllable forces against the uncontrollable
forces of nature. Maneuvering is much more time-critical and less
controllable than bug-correcting, and therefore the flexibility of the means of
cooperation is more important.
This does not mean that maritime work could not benefit from
coordination aids, but they will probably take the form of the "maps" that
more loosely delimits the possible courses of actions. A simple example is
bridge-layout: it is possible that coordination could be made much easier
simply by designing the bridge as a common information space, implying
that the individual instrument is not a one-person instrument but should be
accessible to the whole crew. For example, one could introduce a large
display to which the master could move that information that is crucial in the
present phase of the voyage and which all ought to see clearly.
If we extend the shared field of work to include the sea, we can raise the
question of cooperation between vessels. The simple rule which all vessels
must obey is that no two vessels can occupy the same volume of water. A
volume of water is a scarce resource which only one vessel can possess at a
time. This necessitates cooperation in maneuvering. In order to help the
captain guess the intentions of other vessels, selected instruments of foreign

ships can be made accessible via transponder technology. For example, if
Seal and Atlantic had had access to the rudder angle of Sally Maersk, the
coordination problem in Fragment 5 might have been avoided. Similarly,
access to a selection of the tug-boats instruments might be useful for the
master during berthing, since the tugs are often so close that they cannot be
seen from the bridge, and communication between tugs and pilot is often not
conducted in English. Vessels in fact already try to make their intentions
accessible to other ships: for example, changes of course should be so abrupt
that other ships can perceive them as a course change.
A third possibility we saw at the VTS station at Bremerhafen. It is used
by the pilots to prevent ships passing each other at dangerous bends of the
river Elb.
Ti me
11.00
11.00
10.00
\-..L.___L.....L___L...>..._ Position - - - - - . r - - v - ; : - - - --
10 15 10 25 30 35
Figure 7. Display used by the pilots at the Bremerhafen VTS station
The y-axis represent time and the x-axis position along the river (in the
actual systems, tonnes were used as units of measurements). The vectors
represent the ship's future positions extrapolated by means of the present
speed, and the shaded areas represent the sections of the river where
encounters are prohibited. Vectors pointing left are ships sailing upstream,
those to the right are downstream ships. The simple rule is that vectors
crossing each other inside the shaded areas are prohibited, so band c must be
warned of each other, but no one else.
The Bremerhafen system is a good example of a map overlaid with a
simple protocol. The map asserts true statements about the field of work by
plotting vessels into a grid defined by position X time. The protocol is the
modal element in the display, viz. the vertical columns denoting forbidden
time-space areas.
6. REQUIREMENTS FOR COOPERATIVE WORK
In this section we shall collect requirements for means of cooperation,

irrespective of the medium in which they are realized. The requirements are
extracted from our two field studies, and we shall indicate how they are be
realized in the oral medium and in persistent artifacts.
There are four requirements:
I. They must reduce complexity by dividing information into focus and
background and by supporting standardized protocols.
2. The protocols must be malleable.
3. They must support focus-shifts.
4. They must provide an overview ofthe state of the field of work and the
cooperative arrangement
6.1 Reduction of complexity
The debugging process involved five different roles that would be

assumed by different actors at specific points in time. The paper-based bug
report form reduced the coordination complexity by stipulating the task
responsibilities of each role, the classification of bugs, the intricate flow of
the bug report form, etc.
Forms, checklists, and other types of artifact-based coordinative
constructs enable a reduction of the coordination workload in that they
represent a pre-computation of the central aspects of the coordination work.
The bug report form can be described as a semi-structured form-a
standardized format presenting the roles (actors) with captioned fields to be
filled in, boxes to be ticked, etc.
In the oral maritime communication, the sentential schema constitute the
structural basis for the elaborate backgrounding of coordination information
reducing complexity exponentially, the backgrounded slots being the oral
versions of pre-computation. The schema represents a finite set of place
holders without which one would have to cope with a much wider spectrum
of information. Schema II is the sentential schema underlying the
interpretation of a oral rudder command. In this case only the information in
bold face has focus (is uttered), the rest of the paradigms remaining in the
background.

Helmsman should turn The wheel Port twenty now
Schema 11. Sentential schema and coordination mechanism

The bug-form also uses the technique of backgrounding since it too

requires the designer to only address a few focus paradigms at each time
step. However, the backgrounding is done differently. Whereas oral
coordination often does not express the backgrounded information,
coordination mechanisms display it explicitly. For example, in the bug-form
the identity of the component to be corrected is not focused in the
subsequent Work Distribution, but it is still explicitly represented in the
form.
6.2 Malleable protocols
If protocols are not implemented in a form that is causally active, i.e. as a

computer program, they must be continually maintained by the actors in
order to stay in business. On the one hand, reproduction makes them
vulnerable to errors, but, on the other hand, reproduction enables changes
too, which means that protocols can evolve and be adapted to new
circumstances. Oral coordination accomplishes this task be allowing self-
reference, i.e. one can speak about the protocols by means of the protocols
themselves (Fragment 7-8). The bug-form is not self-referential, since it
contains no entries for changing the procedure of bug-correcting. In order to
do this, the designers will have to resort to varieties of language that allow
for discussing changes of cooperation. There are in fact examples of self-
referential form-based protocols. In a study of the usage of a change request
form in a large manufacturing company (analyzed as a coordination
mechanism), Hans Andersen reports that the request form itself several times
were used to change the protocol according to which the form should be
used (HHK Andersen, 1997). Inventing self-referential coordination
mechanisms is therefore necessary if the mechanisms are to retain the
malleability inherent in oral communication (see Stiemeriing, & Cremers
1998: 303 for attempts in this direction).
6.3 Focus-shifts.
Since reduction of complexity entails that some information must be

backgrounded, and since very few kinds of work - if any - are completely
predictable, coordination mechanisms must support focus-shifts where
backgrounded and presupposed information is re-focused. Focus-shifts are
required in any kind of work (B0dker 1996) and coordination work is no
exception.
Oral communication allows for smooth focus-shifts, as was evident in
Fragment 5 that also provides the reason why focus-shifts are indispensable:
cooperating people will misunderstand one another and therefore need
methods for discussing and changing tacit assumptions. The bug-report did
not support focus-shifts except in the sequence in which fields appear on the
paper, and in that the actors can choose to use the fields for describing other
aspects than stipulated.
6.4 Overview over the field of work
A major benefit of the bug-form is that it makes it easier for the

developers to maintain an updated overview over the field of work: the
information needed is in the same format and in the same place. In maritime
operations, an updated overview of the state of affairs is just as important,
but it is realized by people verbalizing what they do and see. Information
which all need is routed via the intercom with loudspeakers in most
locations: an example is announcement of the time when clocks should be
reset because of a different time-zone; another the call of crew half an hour
before their watch. Walkie-talkies are used to coordinate actions in bow and
stern during berthing.
6.S Dimensions of coordination methods
From the two cases-studies we can extract four dimensions relevant for
coordination mechanisms. Two of them are pragmatic since they relate to the
medium: we distinguish between Persistent vs. Non-persistent media, and
Passive vs. Active media.
By a persistent medium we mean a medium that maintains its
information over time. In non-persistent media, information is lost and must
be recorded elsewhere. The bug-report is persistent whereas oral
communication is non-persistent.
By passive media we mean media that cannot cause actions to happen by
themselves, whereas active media can do this without human intervention.
They are executable on some machine. Both oral language and the bug-
report are passive media since both need human intervention, whereas the
VTS system in Fig. 9 is active since the information about vessel positions
changes without human intervention.
However, these two dimensions are not sufficient, since there is are
differences between the VTS system and the computer-based version of the
bug-report form outlined in Carstensen (1996). One difference is that the
computerized bug-report contains a facility for automatically displaying the
next step of the protocol with a default receiver filled in. Thus it explicitly
refers to the next step in the work process, whereas the VTS system only
indicates possible problematic situations, not the methods for handling them.
Another difference is that the state displayed by the VTS system is the
state of the field of work - vessel positions in the Elb river - whereas the
state displayed by the bug-report system is not the field of work, but the state
of the work· arrangement. The computerized bug-report does not provide
access to the software system under construction, only to the allocation of
tasks to actors and the timing of tasks.
In order to capture these differences we need to look at the semantics of
the representations: do they primarily refer to a State or a Process, and do
they primarily refer to the Work Arrangement or to the Field of Work?
The dimensions should be taken as continua, but sometimes one of the
oppositions dominate. For example, the VTS system mainly denotes a state
and mainly refers to the field of work, with the stipulated coordinative tasks
superimposed in the form of the shaded areas.
This tendency to assign a superordinate, structuring role to one of the
terms can be re-found in conventional text genres. In descriptive genres, the
state provides the structuring, and actions are only mentioned in connection
with objects or locations. In narrative genres, things are opposite: here
actions provide the general structure, and objects and locations are only
entered when motivated by actions. In the former case we shall talk about
maps, in the latter about scripts (cf. a similar notion in Schmidt 1999).
Let us now see how our examples can be characterized by means of these
four dimensions:
Table 1. Classification of various means of coordination

Pragmatics Semantics
Paper-based Bug Report +Persistent, + Passive + Work Arr, + State.
Computer-based Bug Report +Persistent, + Active + Work Arr, + State, + Process.
VTS system +Persistent, + Active + Fld of Work, + State
Object reporting -Persistent, + Passive + Fld of Work, + State
Worker reporting -Persistent, + Passive + Work Arr, + State
Work distribution -Persistent, + Passive + Work Arr, + Process
The paper-based Bug Report is classified as +State, since the protocol is

only weakly mentioned in the form. In opposition to this, the computer-
based version is [+ State, + Process], since the screen displays the current
state of the cooperation as well as stipulations for future actions. The VTS
system is a map of the Field of Work overlaid with information pertaining to
cooperation [+ Fld of Work, + State].
If we look at the different communicative functions, they can also be
classified according to the four dimensions. All share the features [-
Persistent, + Passive], but whereas Object reporting (has it been done to it?)
is [+ Fld of Work, + State] since it concerns the state of a work object,
Worker Reporting (has she done it?) is [+ Work Arr, + State] since the topic
is the status of the tasks of a worker. Finally, Work Distribution, (who

should begin doing it when?), is [+ Work Arr, + Process] since it concerns
the assignment of a task to a worker.
Some of the concepts introduced above can be defined in these terms. For
example, a protocol is defined by the reference of its representations: [+
Work Arrangement, + Process]. An executable protocol is a protocol with
particular pragmatic properties: [+Persistent, + Active]. A Map is a
representation in which the State is the structuring factor, and the Process
must be inferred, whereas a Script is the opposite.
Finally, let us return to the concept of a Coordination Mechanism defined
in Section 4.4. We can now define it as possessing the features [+Persistent,
+ Work Arr, +Process], so the computerized Bug Report is a coordination
mechanism, whereas the VTS system is not.
We propose the term Means of Coordination for any representation that
is used for coordinating actions in a common field of work. This is a
pragmatic definition, referring to the use of the representation. Within this
broad class of representations, the Coordination Mechanism is a special
means of coordination that satisfy further pragmatic and semantic
restrictions: it must be persistent and refer to features of the work
arrangement.
7. ACKNOWLEDGEMENTS
This research documented in the paper could not have been conducted
without the invaluable help of numerous people at Mrersk Seal and and Foss
Electric. We would like to thank the hospitable people at A. P. Moller
Copenhagen for hosting us, and we thank the crew of the Sally Mrersk for
their considerable support during the field study. We thank Steven Verjans
for useful comments. The research has partially been funded by the Center
for Human-Machine Interaction (The Danish Basic Research Foundation)
and the DIWA project (The Danish National Research Foundation). All
errors naturally remain the responsibility of the authors.
REFERENCES
Andersen, Hans H. K. , 1997, Cooperative Documentation Production In Engineering Design.
The 'Mechanisms ofInteractions' Perspective, Centre for Cognitive Informatics, Roskilde
University.
Andersen, Peter Bogh, 1997, A theory of computer semiotics. Semiotic approaches to

construction and assessment of computer systems, (Paperback, 1997), Cambridge
University Press, Cambridge.
Bodker, S. , 1996, Understanding Computer Applications in Use - a Human Activity
Analysis. In Signs of Work, eds. B. Holmqvist, P. Bogh Andersen, H. Klein & R. Posner.
Gruyter, Berlin, 325 - 348.
Carstensen, Peter H, 1996, Computer Supported Coordination, Writings in Computer Science
(No. 61), Department of Computer Science, Roskilde University, Roskilde, Denmark.
Carstensen, Peter H., and Carsten Sorensen, 1996, From the social to the systematic.
Mechanisms supporting coordination in design, Computer Supported Cooperative Work.
The Journal of Collaborative Computing, vol. 5, no. 4, pp. 387-413.
Chong, S. & Kecheng Uu, 2000, A semiotic approach to increase the design quality of agent-
based information systems. In, Proc. of the 3rd Workshop of Organizational Semiotics,
Univ. of Stafford.
Falzon, P. , 1983, Understanding a technical language. Institut National de Recherche en
Informatique et en Automatique ..
Falzon, P. , 1984, The analysis and understanding of an operative language, in B. Schackel
(ed.), INTERACT 1988, North-Holland.
Fitzpatrick, Geraldine, Wiiliam J. Tolone, and Simon Kaplan, 1995. Work, Locales and
Distributed Social Worlds, in H. Marmolin, Y. Sundblad, and K. Schmidt (eds.),
Proceedings of the Fourth European Conference on Computer Supported Cooperative
Work - ECSCW'95, 10-14 September, 1995, Stockholm, Sweden, Kluwer Academic
Publishers, pp. 1-17.
Halliday, M. A. K. , 1978, Language as Social Semiotics. The Social Interpretation of
Language and Meaning, Edward Arnold, London.
Harper, R. R., J. A. Hughes, and D. Z. Shapiro, 1991, Harmonious Working and CSCW,
Computer technology and air traffic control, in J. M. Bowers and S. D. Benford (eds.),
Studies in Computer Supported Cooperative Work. Theory, Practice and Design, North-
Holland, Amsterdam etc., pp. 225-234.
Heath, Christian, Marina Jirotka, Paul Luff, and Jon Hindmarsh, 1993, Unpacking
Collaboration, The Interactional Organisation of Trading in a City Dealing Room, in G.
De Michelis, C. Simone, and K. Schmidt (eds.), ECSCW '93. Proceedings of the Third
European Conference on Computer-Supported Cooperative Work, 13-17 September 1993,
Milan, Italy, Kluwer Academic Publishers, Dordrecht, pp. 155-\70.
Helander, Martin, and Mitsou Nagamachi (eds.) , 1992, Design for Manufacturability - A
Systems Approach to Concurrent Engineering and Ergonomics, Taylor & Francis, London.
Holmqvist, B. , 1989, Work-language and perspective, Scandinavian Journal ofInformation
Systems, vol. I, no. l.
Malone, T. W., K. R. Grant, K. -Yo Lai, R. Rao, and D. Rosenblitt, 1987, Semistructured
messages are surprisingly useful for computer-supported coordination, TorS, vol. 5, no. 2,
pp. 115-13l.
Schmidt, Kjeld, 1994, Modes and Mechanisms ofInteraction in Cooperative Work, Riso
National Laboratory, P.O. Box 49, DK-4000 Roskilde, Denmark. [Riso-R-666].
Schmidt, Kjeld, Of maps and scripts, 1999, The status offormal constructs in cooperative
work, Information and Software Technology, vol. 41, pp. 319-329.
Schmidt, Kjeld, and Carla Simone, Coordination Mechanisms, 1996, Towards a Conceptual
Foundation ofCSCW Systems Design, Computer Supported Cooperative Work. The
Journal of Collaborative Computing, vol. 5, no. 2-3, pp. 155-200.
Stiemerling, o. & A. B. Cremers, 1998. Tailorable component architectures for CSCW-

systems. In, PDP '98 (6th Euromicro workshop on parallel and distributed processing),
Madrid, Spain. IEEE Press, 302-308.
Strauss, Anselm, 1985, Work and the Division of Labor, The Sociological Quarterly, vol. 26,
no. I, pp. 1-19.
Chapter 3
Intertextuality at Work: Large Scale Organisation of

W orkpractices
Rodney 1. Clarke
1. INTRODUCTIONS
The relationship between a system and its organisation is considerably

more complex and interrelated than is generally recognised in the
information systems research literature. Within this literature a system is
often construed as a monolithic entity that impacts upon the organisation
with the frequent implication that systems or their features can be 'best fit' to
a given social milieu solely according to technical and economic rationale.
In contrast, semiotics has the potential for providing a range of new theories,
concepts and practices for exploring the complex interrelationships between
systems and organisations. Clarke (1998) applied a combintion of a semiotic
and contextual model of language called systemic functional linguistics or
SFL (Halliday 1978, 1994; Halliday and Hasan 1976, 1985; Martin 1992;
Ventola 1987) and a compatible semiotic theory based on it called social
semiotics (Kress 1985, 1988; Thibault 1991; Lemke 1995) to information
systems analysis. This combination of theories is referred to as systemic
semiotics (Fawcett in Ventola 1987). An unusual aspect of this theory is that
it uses text as the unit of analysis instead of one or more models of the sign.
A systemic semiotic workpractice framework has been developed in order to
understand the relationships between specific system features within their
organisational contexts (Clarke 2000). This framework is applied to the
60 Rodney J. Clarke
study of an end-user developed operational level administrative information

system called Automated Library and Borrowing System (ALABS). The
system was commissioned in 1986 to support the operations of the
Microcomputer Laboratories at the University of Wollongong (Clarke and
Athanasiadis 1987) to primarily process the loan and return of physical items
including software (on disk) hardware, and audio-visual items to users of the
facility. Applying this framework revealed that discrete workpractices
associated with information systems may enter into large-scale workpractice
assemblages that influence the social structure and participant roles, mandate
new or modify existing social activities and actions, and prescribe the
symbolic means by which other workpractices are enacted in organisational
contexts. This paper describes the large-scale textual organisation of
workpractices in ALABS and identifies three classes of intertextual relations
that link discrete workpractices into larger assemblages.
2. TEXTS AND SYSTEM FEATURES - GENRES

AND WORKPRACTICES
Unlike classical information theory that does not recognise the existence
of 'context', systemic semiotic theory provides explicit theoretical
relationships between a text and the contexts in which it is used (Clarke
1995). All kinds of texts, including those associated with work in
organisations, are enacted and negotiated in specific social occasions and
settings that have an important effect on the texts themselves. The term
redounding is used to refer to the means by which the semiotic system of
language realises context and visa versa. Redounding refers to the fact that
language construes social context, as well as being construed by social
context, while subsequently re-construing its social context, and so on.
Martin (1992, 493) argues that" ... texts are social processes and need to be
analysed as manifestations of the culture they . . . in large measure
construct... [and consequently the theory] ... has to take some responsibility
for a theory of contexts in which language plays a part". The anthropologist
Bronislaw Malinowski's concepts of Context of Culture and Context of
Situation form the organising scheme of the linguistic theory of contexts
developed in SFL, see Figure la. This is not surprising given that Michael
Halliday - the developer of SFL - was significantly influenced by the British
linguist J. R. Firth who advocated amongst other things a contextual theory
of meaning and an emphasis on the context of situation in the analysis of
language (Crystal 1985, 122). In the stratified form of SFL developed by
Martin (1992, 1994), this organising scheme is mapped, as indicated by the
grey arrows in Figure I, to two distinct strata referred to as Genre and
Intertextualify at Work: Large Scale Organisation of Workpractices 61
Register respectively. Strata are a major characteristic of the stratified SFL

model utilised in this study. Similar to the stratificational grammar
developed by the American linguist Sydney Lamb, from whom the approach
has been adapted by Martin, each distinct stratum deals with the modeling of
a specific "... aspect of linguistic structure which can be stated
independently of the structures operating at other strata" (Crystal 1985, 288).
The Genre stratum involves the kind of conventional patterns in text, which
are recognisable in particular cultural contexts and account for meanings
associated with the Context of Culture. Here 'culture' is usefully interpreted
to include organisational as well as national culture. The Register stratum
assists in realising a particular genre by providing specific information about
a given situation. It accounts for the meanings associated with the Context of
Situation, see Figure 1b.
A text always includes some aspects of the situation in order for it to be
understandable. Register describes how the immediate situational context of
the language event affects language use. Register consists of three
identifiable 'variables' (Hasan in Halliday and Hasan 1985) referred to as
field, tenor and mode, which must be considered together. Field can be
glossed as the topic or focus of the activity about which the text is
concerned- its social activities and actions. Mode can be glossed as the
communication channel used in the text; speech or writing for example. The
specific constituents that make up field, tenor, and mode and the methods for
recovering them from texts are described elsewhere (Martin 1992; Clarke
2000). The category of genre in SFL has been particularly concerned with
identifying in texts, the "staged goal oriented social processes which
integrate field, mode and tenor choices in predictable ways" (Halliday and
Martin 1993, 36). Social occasions are always conventional to a greater or
lesser degree, and therefore produce conventionalised forms of texts.
Examples of conventional texts include memos in organisations, essays at
universities, as well as specific routines in the form of work language
patterns (Holmqvist and Andersen 1987) associated with the use of
information systems. A study by Clarke (1996) conducted over the period of
a decade, revealed that one of the two major models of genre developed
within SFL (after Hasan in Halliday and Hasan 1985) could not directly
account for the diachronic changes evident in successive versions of ALABS
systems features. There have been four versions of this system to date.
Martin's (1992) model of genre contains several desirable features including
a systemic theorisation of genre, which allows for the existence of canonical
families of related or agnate genres. However, both Hasan's (in Halliday and
Hasan 1985) and Martin's (1992) approaches rely on the existence of what
the author has referred to as the obligatory-optional element distinction
(Clarke 1996). This concept implies that a text must have a full complement
62 Rodney J. Clarke
of genre defining obligatory elements in order to be considered as a member

of a given genre. While this concept seems reasonable, it has been
questioned in several studies (Ventola 1987; Clarke 1996). The concept of
genre advocated here utilises theoretical affinities between systemic
functional linguistics and social semiotics especially the work of Bakhtin (in
Todorov 1984), Althusser (1971), Foucault (1972), and de Certeau (1984).
This is a suitable combination of theories given the historical relationship
between them (Lemke 1995; Fairclough 1992; Hodge and Kress 1988; Kress
1985; 1988) and the practical importance of combining them (Fawcett in
Ventola 1987). Having briefly introduced some relevant concepts in
systemic semiotics (emphasising those in SFL), these will be applied in an
attempt to understand the interrelationships between system features
associated with workpractices in their organisational contexts.
(a) (b)
...-context of- " Genre
(;
Culture
/C~ ~egister
Situation
Text Language
Figure 1. (a) Malinowski's Context of Culture and Situation are theorised as (b) the bi-stratal
organisation of context within systemic functional linguistics (based on Martin 1994). Grey
arrows indicate the mapping between the two.
When work is conducted in organisations, texts are always involved

either directly or indirectly. This can be seen to occur even in manufacturing
contexts, which would under most circumstances appear to be dominated by
action, see Clarke (1999; 2000) for further clarification. Work in
organisations can be thought of as comprising patterns of communication
and action. The term workpractice is used to refer to these patterns in work
and is defined operationally in terms of one or more text types and zero or
Intertextuality at Work: Large Scale Organisation of Workpractices 63
more action types. When the work being enacted utilises an information
system, then the constituent system features being used to perform the work
are also negotiated through the enactment of workpractices. The relationship
between systems features, workpractices and organisations proposed here
stands in contrast to at least one major strand of thinking within the
'organisational impact' literature which would view system features as
directly producing unmediated social effects with respect to organisational
processes. Using the systemic semiotic approach advocated here, a system
feature is described by gathering texts associated with its use. Systems
features associated with operational level systems often involve the
negotiation of goods and services and so are often realised through spoken
language texts. Common examples include borrowing books from libraries,
paying land rates at the council chambers, and providing technical advice to
computer users from help desks. Many actual text instances need to be
collected in order to understand the variety of ways in which a particular
system feature can be realised using spoken language. Each spoken language
text needs to be recorded and transcribed. Of course, many system features
in operational level systems generate written language texts, including
forms, reports and other types of documents. For the most part, we confine
our discussion to considering spoken language texts in the form of service
encounters conducted using ALABS at the Microcomputer Laboratories,
University of Wollongong. The ALABS system is shown in Figure 2a. Also
provided are images of the physical facility. In its current location and
configuration (Building 40) the facility consists of seven PC laboratories of
20 machines each. An example of one of the laboratories is shown in Figure
2b. Figure 2c shows an image of the office window through which service
encounters are performed. Figure 2d shows the main corridor of the facility
providing access to most of the laboratories.
From a systemic semiotic perspective, ALABS and other information
systems consist of two sets of system features which may have
workpractices associated with them. The first set of system features
comprises those operations required to support the technical infrastructure
for example packing databases. These system features may have texts
associated with them- called indirect texts- that the analyst will need to
identify along with the stakeholders responsible for this aspect of the
technical infrastructure. For example, a Microcomputer Laboratories staff
member called the Laboratory Assistant was responsible for 'packing' - that
is compressing and archiving- the ALABS Sessional Transaction Database
generated as a consequence of ALABS Student Loan and Return activities.
However, a meeting was required between the Laboratory Assistant,
Programmer, and Operations Supervisor to determine the manner and timing
of this operation. This meeting constituted an indirect text- one that is
64 Rodney J. Clarke
obviously not supported by the information system itself. The second set of
systems features comprises operations required to support organisational
workpractices. The texts associated with organisational workpractices are the
justification for the existence of those texts associated with the technical
infrastructure. The subsequent discussion of ALABS systems features
focuses on organisational workpractices. These system features are realised
in the form of direct texts. An analyst requires no elicitation techniques to
uncover these workpractices, which are self-evident in the workplace. The
ALABS Student Loan is an example of a direct text.
(a) (b)
(c) (d)
Figure 2. Images from a QuickTime Virtual Reality (circa 1995) of the Microcomputer
Laboratories physical facility at its current location (Building 40), where (a) shows inside the
office area. The PC on the right hand side runs ALABS while the others are operating as
Novell Servers. Image (b) ·shows the interior of the seven teaching laboratories, (c) shows the
exterior of the office area, note that the window in the foreground is where service encounters
take place, and (d) is a view down the main corridor of the facility .
Our own experience of purpose-oriented organisational discourse,

especially the kind of communication associated with information systems
use, is that it is goal directed and structured into patterns that have a finite
and generally predictable range of forms. SFL methods can be used to
identify the staging of texts including workpractice texts associated with
system features. Each stage in a text is referred to as a genre element. The
specific sequence of genre elements (syntagm) identified in a workpractice
text is referred to here as a genre sequence. In order to characterise the
variety of textual realisations that are typically associated with the enactment
of a given workpractice (paradigm), many texts associated with the
workpractice are examined. The corresponding genre sequences are merged
together to form a so-called genre digraph that graphically represents the
variety of ways in which a system feature was or is likely to be enacted
(Clarke 2000).
The method is exemplified using a transcript of an actual Student Loan
conducted using Version 2 of ALABS. The first column in Table 2 indicates
which participant holds the current tum (that is who has the 'floor' either
verbally and/or non-verbally). An 'L' indicates that the current tum is being
taken by the Microcomputer Laboratories Staff Member, responsible for
entering details of the loan to ALABS, while an'S' indicates that the current
tum is being taken by a student (in this particular case one who was
unfamiliar with the system). The second column provided the so-called
mainlines which represent what was said and done. The term 'mainline' is
the name used for this part of a transcript in the Codes for the Human
Analysis of Transcripts CHAT transcnptlOn and coding system
(Mac Whinney 1995). The code for a specific genre element is provided in
the third column that represents a particular functional stage in the work of
conducting a Student Loan. The tum number is shown in the fourth column
of the table. It is worth pointing out that not all Student Loan texts are this
lengthy. Students and Staff who are familiar with the system can conduct a
satisfactory Student Loan in a matter of seconds. For Student Loans, almost
all of the stages can be conducted non-verbally and only one word needs to
be spoken- the lexical item that identifies an item being requested by the
student. The particular example in Table 2 has been selected because it
possesses almost all of the elements that are recognised as belonging to this
genre. The genre element codes in the third column of Table 2, are described
in more detail in Table 1, a list of identified elements associated with the
Student Loan genre. This kind of table is referred to as a genre element
inventory; it lists an element code, its full name and its function. Note that
the transcript in Table 2 does not use the full complement of elements
available in the genre element inventory. The phatic element Greeting was
apparent in other Student Loan transcripts. An atypical aspect of this
transcript is that it includes Regulations and Enrolment elements used for
registering a Student on ALABS. This was only required once per session
66 Rodney J. Clarke
per student. It also required that the student read and signed a contract with
the Laboratories called the Conditions of Loan form, which amongst other
things required the student to return all items in working order in a timely
fashion. Grey areas in Table 2 indicate that the Student is wandering away
from the purpose of the workpractice. The Labstaff member attempts to
quickly reorient the student back into the genre.
Table 1. ALABS Student Loan Element Inventory
Code Element Function

G Greeting phatic initiation
SR Service Request request for loan services
IS Identification Sought student-id or equivalent retained
RE R~J(ulations student reJ(ulations
E Enrolment enrolment of student requestor
MO Materials Out items provided
F Finis phatic conclusion
The genre sequence for the transcript in Table 2 is provided in Figure 3a-
the directed graph notation is after Clarke (2000). The triangle on the left
signifies the beginning of a genre sequence or directed graph, Genre
elements are indicated by circles and include thelabel which uniquely
identifies it. Arrows indicate sequence. The inverted triangle signifies the
end of the sequence. Note that no distinction is recognised or made between
elements that are 'obligatory' or those that are 'optional'. Single genre
sequences like that shown in Figure 3a are not normally provided unless it is
being used for comparison purposes. Although not shown here, a
comparison chart was developed in order to show the positional matches or
mismatches between elements in sequences or digraphs. This single
sequence is included here to show the contribution it makes to the genre
itself, shown in Figure 3b. Other genre sequences have been merged together
to form the genre digraph. The elements that are not grey were contributions
made to the genre digraph by other transcripts.
3. ALABS WORKPRACTICES INVOLVING

STUDENTS AND LABSTAFF
As previously mentioned the primary purpose for ALABS was the loan
and return of physical items to Students and Staff. We will describe only
Table 2 Transcript of a Student Loan (double column layout) conducted using ALABS
Version 2 (after Clarke 1996, 2000). Genre elements codes are provided in Table 1, and are
separated by thick black lines. A description of the transcript and an explanation of its other
conventions are provided in the text.
L wnat senware ao you wan S Don't tell me I have to do this every 31

:> I aon [KnOW • time I want to get a disk
L Wntcr sUDject are you enrollea In. • L NO, you OnlY nave 10 ao mls once at ••
:> m not sure • the beginning of each session
L Uk, wno'S ma lecturer 0 :> Wnat nappens WI aom sIgn ~. ••
:> un, I alan [ eaten me name
wnlcn uepamnenr!
0 L
:>
en you aon ge any SOJIWare
1:>luaem reaas aocumem ana sIgns
••
L
S
L
Commerce
NO ,"a s a acu~y, not a aepa mem! .... 8
•
10
L
(;an [ nave ma alSk now
NO, ave 0 en er your 08181 s
[Staff takes the form and the pen]
c
'0
""
5 un, Inform a lon.Sys ems
L s • a Irs year course [Staff enters details to AlABS]
S Yeh 12 :> NOW can I nave tne aISK. JI
L HOW aoes "'1,,'" s0Ul1Cl"?
,.
~ L Yes, Out YOU' II neea one Of mese 38
:> ean, I tnlnk mars II [Staff member pickS out a lab pass,
L Ne you eamlng "",,,I,, , 15 and swipes ~ with the Barcode wand]
:;
L
Tean we re eamlng programmIng
UK, you neea a uuti alsk.
'0
17
:;
L
iVnars mar(
Irs a lao pass. It allOWs you to go Into
LV
••
40
Can I have your Student Card? a particular lab and usa the machine
:; I aon'll ave I 1a w~h the same number as on the card.
L we can'l give you any senware I. All the machines are numbered.
.
unless you have your student [Staff hands lab pass to student]
identification card l"lBn memcer anos over SOTlWare mu
S [Looks through wallet] IS :> :m~~:",pens W want· 0 go nave a
Oh, yeah I've got ~ 20
L All ngnt, give n to me please 2' L ~Ine, mat·s Uk. Just make sure yoU're 42
:; I~,uClllnt nanos over sluaent in the right laboratory. When you want
identification card] to leave make sure thet you bring both
00 I get ~ back 22 the disk and the Labpass back.
YOU get U18 cara DSCK wnen you gIve
..
••
L Otherwise you don't get your card and
us the software back you don't get to borrow anything else.
:> ... an I laKe me SOJIWare nome Ok? That is for PC Lab 2, there's a
L NO, you can onlY use me SOJIWare In .0 map on the notice board, tells you
one of our three PC labs where~ is.
[Lab Staff swipes the student card
with the barcode wand] "L ...an go 10 any otnar IBO.
:;ura DUI you can only WOrk In Lao 2,
••
44
~~l!'~:; nas no recora 01 me sluaenlj Ok? The pess is for Lab 2; you must
[Lab Staff must enter the Student we can renew ~ for you if you need
details into the ALABS database] more time.
L ~ your,!'rst."me nare .8 :; HOW ao renew Ina senware 45
:; ea <, L JUst Dnng me sonware ana Laopess 10
L I nougnl so, you ve goilO 1111 oul ona 28 back to us and we can renew the loan
of these forms for you, Ok?
[Hands over the Rules and :; wnare Lao 2"f 4,
Regulations for using the L I "" me "''''' are on Ul e map on UlIiII '0
LaboratOries, students must sign] notice board. Lab 2 is across the foyer
:;. ~rs tnIS.aD~7 He 2. on the right-hand side, OK?
L It's the rules and regulations for 30 5 Ok, tnanks. F 49
these
Laboratories. Read them and sign
down the bottom.
[Staff member hands over the pen]
those workpractices evident as texts occurring between Students and

Labstaff, excluding from the discussion other types of participants and their
associated workpractices. With the exception of the Conditions of Loan
Form, all workpractices involving Student and Labstaff were spoken
language texts conforming to variations on the service encounter genre used
when negotiating goods and services. The order of development of these
system features and the interrelationships between associated workpractices
over time is deferred until the next section.
68 Rodney J. Clarke
(a) IS-
(b)
Figure 3. Relationship between Genre Sequences and Digraphs and their notations during
systemic semiotic analysis (after Clarke 1998, 2000). The sequence of genre elements from
the ALABS Student Loan text provided in Table 1 is shown as a genre sequence in (a). The
Student Loan Genre represented by a genre digraph is derived by merging the Genre
Sequences of many Student Loan texts is shown in (b). The genre elements from the sequence
in (a) are shown in grey in (b).
3.1 Conditions of Loan Form
When ALABS was commissioned, there was no campus wide policy

pertaining to student usage of university computing facilities. The Labstaff
drafted a list of requirements and regulations, which became known as the
Conditions of Loan form. It was based on a Loans Form that has been
developed in a previous manual loan system. Students seeking to use the
Microcomputer Laboratories were required to sign this form prior to them
being placed on the system. As the student database was purged every
session, students would sign the Conditions of Loan form every session.
These forms were stored at the Laboratories. Students who failed to return
software or other material would have their accounts frozen until such time
as the materials were returned. Students were left with little doubt as to their
duties and responsibilities in relation to borrowing items from the facility
(see Clarke 2000). This was reinforced by the use in this text of particular
lexical resources (in the form of legallexis), and the deployment of specific
grammatical resources associated with the degree of certainty in the
exchange of information (modality) and the imposition of obligations
(modulation). Also, the specific register features strongly indicated that an
unequal social relation of power existed between Students and the
representatives of the facility. In addition, the generic organisation of the text
itself was in the form of a contractual regulatory genre.
3.2 Student Loan
Students request the loan of items from the Microcomputer Laboratories.

The items may be in the form of software, manuals, or hardware. A Labstaff
member uses the ALABS Student Loan feature to record items that are
loaned to students. The primary social goal of this workpractice was the
correct recording of the loan in order to maintain the integrity of the holdings
and to not infringe the licensing agreements with the software distributors. A
secondary goal was the efficient loan of items to Students. For an example of
this workpractice, see the genre element inventory in Table 1, an example
transcript in Table 2, its representation as a genre sequence in Figure 3a, and
its contribution to the Student Loan genre provided in the genre digraph of
Figure 3b.
3.3 Student Return
Students must return any and all items borrowed from the Microcomputer
Laboratories. Those who failed to return items would be automatically
placed on an Offence database and precluded from borrowing further items.
A Labstaff member uses the ALABS Student Return feature to record the
return of items. The primary social goal of this workpractice was to correctly
record the details of the return in order to maintain the integrity of the
holdings and to not infringe the licensing agreements the facility had entered
into with the software distributors. Related to this goal was the need to be
able to establish an accurate audit trail for the Loan. A secondary goal was
the efficient processing of Returns.
3.4 Student Append
This workpractice perhaps more than any other in ALABS demonstrates

how systems features can be considered as repositories for patterns of human
communication that can form the basis of new system features. The purpose
of a Student Append was to provide students with the ability to borrow
additional software items while still having other items on active loan. For
example, a Student may wish to borrow word-processing software to
complete the documentation for a programming assignment. Students were
able to conduct Student Appends only after they had already undertaken a
Student Loan. Because Students had a currently active Student Loan, the
Microcomputer Laboratories were already holding their student
identification card. The Student needed to simply request additional items.
Even though the Microcomputer Laboratories were already holding the
Student Identification Card, Student Appends did not always proceed
70 Rodney J. Clarke
smoothly. Often the photographs on the student identification cards were of

a poor quality. On occasion the Labstaff member would ask for the students
name to help locate the card, or alternatively the student would lean over and
point out their card. Interestingly enough, a Student Append was conducted
with much less formality than a Student Loan. A number of factors acted to
reduce the unequal power continuum compared to Student Loan situations.
These included the fact that students needed to be actively involved in their
own identification and had already gone through the process of identifying
themselves and so were already complying with the laboratory procedures.
3.5 Student Loan Renewals
The Student Loan Renewal feature was introduced to permit students to

renew items already loaned to them which had become overdue, or were
about to become overdue. The default length of time for an item loaned to a
student was two hours. During high usage periods, the demand for specific
software items could be acute. This was particularly the case when
assignment deadlines were looming for large subjects. So that students could
potentially work for longer than two contiguous hours, while also preventing
attempts to monopolise software items, students were expected to undergo
Student Loan Renewals.
Labstaff were required to follow a set of daily, weekly, sessional and
yearly procedures to ensure the correct functioning of ALABS. Some
procedures were technical; others directly associated with workpractices.
Daily procedures required that Trainees and the Laboratory Assistant
regularly "[t]our the laboratories ... [and] in peak demand periods ensure that
no student spends in excess of two hours on a computer" (Clarke et al 1989,
v). During peak periods, Labstaff would select the Overdue Student Loans
feature to determine who was late. They would then locate the Student
Identification Cards for these late borrowers, tour the Laboratories to find
these students. Students were requested to either renew their loan or return
all borrowed items and leave the facility. Student Renewals could be granted
to those students who were late and who requested one, when approached by
Labstaff in a laboratory during peak periods of activity. In this case, Overdue
students were asked to accompany the Labstaff member to the Service Desk,
where the borrowed items would be reentered into ALABS. The student
would then return to the Laboratories and continue working with these items.
In practice, if the items borrowed by a student were in demand, then any
extensions to loans involving scarce items could be refused. Students refused
the option of a Student Renewal were usually required to immediately
undertake a Student Returns service encounter, as specified in one of the
clauses of the Conditions of Loan form. Only when demand in the
Laboratories was low, would a student who was late be granted a Student
Renewal.
3.6 Student Move
Student Moves were required when students had to be relocated from one
laboratory to another in the facility. This situation occurred if Students were
still within their current loan period, but their laboratory was about to be
used for a Supervised Laboratory Time, and space in another appropriate
Laboratory was available. As the facility supported a variety of platforms, an
appropriate Laboratory was one in which the same software could be used.
Students 'kicked out' of a laboratory by a Tutor would end up back at the
Service Desk to either enter into a Student Return service encounter or a
Student Move service encounter. To conduct a Student Move, the Labstaff
would select the Swap Lab Pass feature of ALABS. They would locate the
student's identification card previously stored from an initial Student Loan,
and then rescan and retain it. They would then retrieve the students current
Labpass, scan and retain it, find a new Labpass for a different laboratory
consisting of the same machines as the previous laboratory, and then scan
the new Labpass and provide it to the student.
3.7 Student Booking
As with Student Moves, Student Bookings were also commissioned in

ALABS Versions 2 and 3 as a response to growing demands on the facility.
The demands for Student Loans were considerable. Student Bookings in
effect provided students with a guarantee of a reserved machine in a
laboratory for a specific time slot. As Student Bookings placed considerable
stress on the operations of the facility, they were only provided during
known periods of high demand, for example when the assignments for large
classes were due. If bookings were in effect, Students were permitted to
book machines one day in advance on weekdays and two days in advance on
weekends. The system required the Student's barcode. If the student
requesting the booking was not enrolled on the system, the student was
enrolled as per the Student Loan workpractices. This was the only other
workpractice to have enrolment elements. It provides a glimpse into the time
pressures and frayed nerves that surrounded this workpractice that it was less
stressful to Students and Labstaff to immediately enroll rather than to enter
into an unwanted Student Loan in order for the student to be registered on
ALABS. The inclusion of enrollment elements also indicates that a
significant number of students did not work outside of their normal tutorial
laboratory times, until their assignments were due!
72 Rodney J Clarke
4. DIACHRONIC STRUCTURAL CHANGE IN

SYSTEMS FEATURES
It was evident that ALABS workpractices changed over time. A

longitudinal study of genres of several 'core' ALABS features (Clarke 1996)
revealed interdependencies between genres associated with related
workpractices. Due to space restrictions we limit the scope of the discussion
to specific examples of direct structural change evident in workpractices
relating Student who borrow and hopefully return items, and the Labstaff
that were responsible for these transactions. Excluded from further
consideration are any so-called non-structural diachronic changes (Clarke
2000, 259), for example the reuse of ALABS system features for other
purposes, and the development and use of redundant functionality. This
section describes the direct structural changes over time for ALABS
workpractices involving Students and Labstaff. These changes suggest the
existence of large-scale structure in the organisation in ALABS Student and
Labstaff workpractices.
Successive versions of the same system feature exhibit structural
differences and are actively altered over time by end-user developers. Three
kinds of structural operations have been used to develop the phylogenetic
account of ALABS workpractice genres provided in this dissertation. Single
elements have been added to genres- referred to as element pasting, or
removed from genres- referred to element cutting. In some cases genres have
been subject to the addition and removal of elements- referred to as generic
elaboration. The Student Loan/Return workpractice provides examples of all
three. Version 1 of the Student Loan Genre was modified in Version 2 by the
inclusion of a single element (Labpass Out), an example of element pasting.
This simple structural change enabled the Labstaff to allocate machine
resources and to manage student loads within the facility. However, by the
time Version 4 of ALABS was released, the complete networking of the
laboratories had already taken place. This meant that the vast majority of
students could simply walk into an available laboratory and select a
computer application from a network server rather than borrowing disks
using a service encounter. Consequently, the Labpass Out element was
deleted from the Student Loan genre because it was no longer needed, an
example of element cutting. Similar structural changes were made to the
Student Return Genre, Version 2 of which saw the inclusion of a single
element (Labpass In), and its subsequent removal in Version 4. Both the
Student Loan and Return genres were subject to several episodes of generic
elaboration across successive versions of ALABS. Significantly, the genres
that comprise this workpractice are referred to as structural complements
because a structural change made to one of them necessitates a
corresponding structural change be made to the other. While not previously

identified in the systemic functional linguistics literature, these structural
complements demonstrate the existence of synchronic and diachronic
interdependencies that can exist between genre. Changes to workpractice
genres could always be explained by reference to changes in the immediate
conditions of the Microcomputer Laboratories, or by changes in the broader
organisational context that had a direct affect on the functioning of the unit.
Changes to ALABS workpractices were not simply limited to structural
operations on single genres, but also included various kinds of modifications
occurring at a variety of different scales, see Table 3. This was not so much a
question about the existence of other 'larger' diachronic units, as it was a
question of 'granularity' or how simple generic operations involving small
units (genre and qualitative elements) could affect extensive changes to an
entire information system. There are several examples where copying entire
sequences of elements expanded the functionality of ALABS. One case
involves the use of sequence copying to develop a Student Append feature
from the Student Loan of Version 1. No example of elaboration at the level
of the genre element was identified, although entire sequences were
elaborated to achieve new functionality. Other examples of sequence
copying were found in ALABS workpractices including Tutor Loans and
Returns and Long Term Staff Loan and Returns. Sequence Cutting was
limited to Class Loans and Returns workpractices.
5. WORKPRACTICE ASSEMBLAGES AND

ASSOCIATIONS
The fact that workpractices like Student Loans/Returns act to constrain

and organise the kinds of modifications made at the level of their constituent
genres was established in the previous section. The mechanism proposed to
account for these direct structural changes was referred to as structural
complements. Clarke (2000) also identified an example of a consequential
structural change in ALABS where a change in a set of workpractices
produced an indirect change in an otherwise unrelated workpractice.
Although, this indirect change is detectable at the level of the genre and
qualitative digraph, no mechanism was described that might account for it.
The small scale or local perspective on workpractices afforded by genres and
qualitative digraphs hampers the articulation of this kind of mechanism.
74 Rodney 1. Clarke
Table 3. Structural Diachronic Changes to ALABS Workpractices involving Students and

Labstaff (based on Clarke 2000). Proxemic and chronemic rationale implies that a change is
due to space considerations or time considerations respectively.
UniUOperation/Workpractice Rationale I Version

I 2/3 4
Elements
Pasting
Student Loans Proxemics ./
Student Returns Proxemics ./
Cutting
Student Loans Proxemics ./
Student Returns Proxemics ./
Sequences * indicates multiple sequences
Sequence Copying
Student Appends ./
Sequence Elaboration
Student Renewal Chronemics ./
Student Move Proxemics ./
Student Booking Chronemics ./
Clarke ( 1996) proposed a new generic structure referred to as an

Assemblage, which comprised groups of genres linked by implicit or explicit
use. Assemblages employ genre associations in order to construct
relationships between genres, which bind together sets of social subjects into
particular configurations of social action, and social relations of power.
Assemblages provide a larger scale or 'regional' perspective on information
systems, and necessitate the search for inter-generic mechanisms at this level
of description. The advantage of Assemblages is that they directly reveal the
social relations of power that bind together sets of social subjects into
textually realised, enacted social workpractices, literally the web of
computing envisaged by Kling and Scachhi (1982). The characteristics of
Assemblages are theorised below. Clarke (2000) identified five different
Assemblages in ALABS. Given that this is a small system, this fact alone
demonstrates the importance of Assemblages and associations in explaining
the end-user development and diachronic changes to this system.
5.1 Semiotics of Assemblages and Associations
Each constituent workpractice within an Assemblage marks out a

component of the available meaning potential as a whole. The concept of
meaning potential refers to the " ... paradigmatic range of semantic choices
that is present in the [semantic system of language], and to which the
members of a culture have access in their language" (Halliday 1978, 109 and
111-112). As a concept it is a 'convenient fiction' in that it would be

impossible to exhaustively describe it for a given Context of Culture, an
entire organisation for example. However, for a specific part of the
prevailing social structure (for example, Labstaff and Students see below), it
is possible to describe the range of options that are characteristic of a
specific set of Contexts of Situation- in effect this is the purpose of
Assemblages.
As Assemblages consist of genres and qualitative digraphs, it seems
reasonable to apply the same characteristics for genres, developed by
Freadman and Macdonald (1992), to Assemblages as a whole. As a
consequence, Assemblages must also be visible and exhibit the
characteristics of persistence and inertia. In terms of visibility, Students
should be aware of the types of workpractices that are available to them and
the rules and regulations that pertain to them. The requirement that
assemblages exhibit persistence and inertia is demonstrated by the fact that
workpractice assemblages are relatively stable over time and capable of
accommodating and responding to changes in the organisation. The social
semiotic claims for genre would seem to be particularly relevant here.
Bakhtin writes that "genre lives in the present, but it always remembers the
past, its beginnings. Genre is the representative of creative memory in the
process of literary [and in this context workpractice] evolution, which is
precisely why genre is capable of guaranteeing the unity and a continuity of
this evolution" (Bakhtin in Todorov 1984, 84). Viewed diachronically,
Assemblages must be seen as being subject to deliberate modification, and
must also be ideological in that they are involved in the patterning and
symbolic arrangement of genres- " ... the more elevated the genre, the more
complex it has become and the more, and the better, it remembers its past"
(Bakhtin in Todorov 1984, 85).
Significantly, Assemblages appear to be generic manifestations of
intertextuality between workpractices. In this case, the workpractices are
associated with an information system. Intertextuality is defined as "[t]he
meaning relations which exist between texts, and the process of making
sense of texts in reference to their relations to other texts" (Schirato and Yell
1996, 239). Systemic Functional Linguistics defines intertextuality in terms
of direct references linking one text to another. Clarke and Mehler (1999)
have employed this usage of the term to describe the generation of
hypertexts from print media corpora. While this definition of intertextuality
as direct reference accounts for some relationships between workpractice
genres that form Assemblages, it cannot account for a large proportion of
workpractice genres in some assemblages. In contrast, social semiotics uses
the term intertextuality to describe "... the ways in which texts embody
meanings that have already been made, in one form or another in other texts"
76 Rodney J Clarke
(Schirato and Yell 1996, 239). These meanings are conventional, requiring
familiarity not intuition. In other words, a user (or reader) understands
specific workpractices (texts) because they have prior experience of them.
The term as it is used within social semiotics is based on Julia Kristeva's
reworking of Bakhtin' s theory of dialogism and heteroglossia (Belsey 1980,
26). The concept of dialogism involves recognising that all texts or
"utterances" function to respond to a prior text and already anticipate a
response. However according to Todorov (1984, 61), " ... not all relations
between utterances are necessarily intertextual. Logical relations must be
excluded from dialogism (for example: negation, deduction, etc.); in
themselves, they do not imply intertextuality (though the latter may be down
to them); the same is obviously true of purely formal, all linguistic relations
in the strict sense (anaphora, parallelism, etc.)". Although the Systemic
Functional Linguistics and Social Semiotics definitions of intertextuality are
mutually exclusive, they both explain different aspects of Assemblages. As a
consequence both definitions are used when interpreting Assemblages and
associations. To disambiguate between these types of intertextuality we refer
to the former as direct intertextual relations and the latter type as dialogic
intertextual relations.
5.2 Types of Intertextuality in the Student LoanlReturns

Assemblage
Assemblages are represented using Assemblage diagrams (Clarke 2000)

that show texts that support organisational workpractices as well as those
that support the technical infrastructure of the system shared by sets of
stakeholders or social subjects. Assemblage diagrams can also be annotated
with the necessary actions needed to trigger the appropriate feature although
this is not done so here. The genre assemblage for the ALABS Student
Loan/Returns Assemblage; see Figure 4. It consists of written genres (ovals
with thin lines), and a series of spoken genres (ovals with thick lines) which
form the core organisational workpractices enacted by Students and Labstaff
using ALABS at the Microcomputer Laboratories. Dashed ovals indicate
that the workpractice was recovered from interviews rather than direct texts-
this is typical of a system like ALABS that is partially decommissioned.
These workpractices involve either negotiating the exchange of goods or the
provision of services in the form of facility access. A key to the abbreviated
names of each genre in Figure 4 is provided in Table 4 which also includes
the full name, the type of text, and the versions in which they were
commissioned and decommissioned. A genre association, indicated by the
thick black lines in Figure 4, joins the Student Loan and Student Return
service encounters to the written regulatory text Conditions of Loan. The
intertextual relations which associate Student Loans and Returns to

Conditions of Loan utilise direct referencing- that is the intertextual relations
recognised in Systemic Functional Linguistics which are referred to here as
direct intertextual relations.
There has been an interesting transformation of the status of the
Conditions of Loan text across various systems at the Microcomputer
Laboratories. The Manual System transformed the informal service
encounters of the Ad-hoc System, into formalised service encounters which
produced a single written text record for each Loan/Return transaction. The
Conditions of Loan were developed at the Laboratories precisely because
there had been no relevant University policy. In the Manual System, a
Conditions of Loan text was an explicit part of every service encounter of
the Student Loan/Return workpractice. But when ALABS Version 1 was
commissioned, the Conditions of Loan text was relegated to being negotiated
the first time a student borrowed items in a given session. At the time
ALABS Version 4 was commissioned, the University's Information
Technology Services Unit had developed a general Code of Practice: Student
Use of Computing Facilities. This enabled the removal of the Regulations
and Enrolment elements of the Student Loan Genre in Version 4. This Code
of Practice was in fact based on the Microcomputer Laboratories Conditions
of Loan text!
In discussing the ideological nature of genres, Thibault (1989, 346)
remarked that genres are "". socially made and used programs for the
making and regulating of certain kinds of meanings and social relations".
This association between Student Loans/Returns and Conditions of Use is an
obvious manifestation of the ideological nature of genres. Its eventual
modification due to the change in campus-wide computing policy exhibits a
shift in 'certain kinds of meaning and social relations'. The modification
removed the burden of the Microcomputer Laboratories in administering and
enforcing compliant behaviour from students. It elevated the social relations
of power between computing service providers and their users from the level
of local practice to that of officially sanctioned campus-wide policy. It also
mandated a common set of expectations, procedures and practices across all
computing service providers and helped to establish the newly formed ITS as
the overarching controlling authority for all computing service providers on
campus including the Microcomputer Laboratories.
78 Rodney J. Clarke
... ---........
( Stud7nt ) ( Student )
....
, -._-...., I.----
r
--
,.".---... ,
Student ) r
" ... --
........
Siudent ) Student
1 ' ...- --'
\ Booking , \ Move ,
' ...- -' _-
\ Renewal ,
, .... -' "-
-- --
_._,- - Loan
\ Append ,
Student LoansIReturns : Intertextual Relations

Direct Elliptical Dialogic
Student Loan Student Append Student Move
Student Return Student Renewal Student Booking
Conditions of Loan
Figure 4. Student Loan/Returns Assemblage with a table showing the types of intertextual
relations that bind genres (based on Clarke 2000, 212 and 259).
Tahle 4. Genres identified as part of Student Loan/Returns Assemblage in Figure 4.
Abbreviation Full Name Text Type Versions

Condition of Loan Conditions of Loan Form written; regulatory 1,4
Student Loan Student Loans spoken; service encounter I
Student Append Student Append spoken; service encounter 1,4
Student Renewal Student Renewal spoken; service encounter 2,4
Stude nt Move Student Move spoken; service encounter 2.4
Student Booking Student Booking spoken; service encounter 2, 4
Student Return Student Return spoken; service encounter I
Offence List Offence List written; operational 2,4
Student List Student List written; operational I
Student Appends constituted a dependent service encounter, one that

required a prior Student Loan service encounter. Although like the Student
Loan it was first released in Version 1, it was developed after the Conditions
of Loan text had been created. Consequently, despite having a very similar
structure to Student Loans, it is not directly referenced by the Conditions of
Loan text. Because no difference in 'lateness' is recognised between late
items borrowed using Student Appends, and those of Student Loans,
Offenses could be generated on either. At first appearance, it would seem
that Student Appends are associated with Conditions of Loan through a
dialogic intertextual relation due to the absence of a direct reference.

However, a logical relationship of succession exists between Student Loans
and Student Appends based on the fact that the latter is dependent on a prior
occurrence of the former. The logical relationship of Student Appends and
Student Renewals to Student Loans, is based on a shared structural
requirement, to have already realised the IS element in the Student Loan and
therefore to already have available the requesting student's Identification
card. Logical relations of this kind are excluded from the social semiotic
definition of an intertextual relation. As a consequence, the term elliptical
intertextual reference, is developed here, and used to refer to this 'omitted'
type of direct intertextual reference. Elliptical intertextual referencing
appears to be the rrieans by which all dependent service encounters are
'locked' into Assemblages. An elliptical intertextual reference also binds the
Student Renewal workpractice, introduced in Version 2/3 to this assemblage.
This dependent service encounter had not been specified at the time the
genre association was first established. Yet it slotted into a growing semiotic
web of intertextuality represented by this assemblage, and was treated as if it
was directly subject to the Conditions of Loan.
Like the Student Append and Renewals, the Student Move workpractice
introduced in Version 2/3 shares the structural requirement of a previously
realised IS element from the Student Loan. However, while it requires a
currently active loan, it is not directly subject to any Offence procedures.
This workpractice is however subject to the normal standing procedures for
Labstaff, students may be requested to move if a class was scheduled for the
laboratory. Therefore dialogic intertextual relations, as defined in Social
Semiotics, link the Student Move workpractice to this Assemblage. Student
Bookings workpractices introduced in Version 2/3 involve the negotiation of
future machine access. Students however were still locked into the normal
standing procedures involved in the surveillance of the laboratories by
Labstaff. Unused machines, even if previously booked by students, were lost
if the student was ten minutes late, or failed to tum up. Student Bookings is
considered apart of this assemblage because it is implicated in the same
intertextuality exhibited by other workpractices.
5.3 Principles of diachronic/ontogenetic development
The Student LoanlRetums Assemblage demonstrates several principles in

the diachronic or ontogenetic development of ALABS. It appears that
Assemblages can't support wildly differing tenor relations within their
constituent workpractices. This characteristic of Assemblages is referred to
here as the principle of the conservation of social relations of power. This
ALABS Assemblage consistently positions Students into specific social
80 Rodney J Clarke
relations of power from which they cannot escape. Each subsequent

workpractice inherits, adopts or amplifies available subject positions (Kress
1988) already established by the Assemblage. This specificity is
demonstrated by looking at the consistency exhibited in the tenor relations
between Labstaff and Students (see Clarke 2000). Also, each newly
introduced workpractice must co-exist in relation to other intertextually
related workpractices within an Assemblage. In terms of the diachronic or
ontogenetic development of ALABS, this is referred to as the principle of
successive activity differentiation. This principle is demonstrated by the
structural similarities evident in the relevant workpractices previously
described.
There appears to be no correlation between the persistence of
workpractices and the type of intertextual references employed in their
associations. For example, one of the ALABS assemblage (Class
Loans/Returns) did not have any direct or elliptical intertextual references at
all, yet it was one of the most complicated assemblages in terms of the
number and interrelationships between its workpractices and also one of the
most persistent. However, there appears to an interesting relationship
between the type of intertextual reference employed in associations and the
ontogenetic sequence of development for workpractices in their respective
assemblages. For those assemblages that were subject to ontogenetic
accretion of new workpractices (Student Loans/Returns and another called
Staff Loans/Returns which was not considered here), the initial
workpractices employed direct intertextual references in their associations.
Once those intertextual relations were established, additional related
workpractices utilised elliptical intertextual relations. Finally, dialogic
intertextuality was used to associate additional workpractices. Clarke (2000)
showed that Assemblages, which did not have any ontogenetic development
in the sequences that made up their constituent workpractices, consisted
entirely of direct intertextual relations in the respective associations. This is
not surprising in the case of one of assemblage because it was related to
ALABS Management functions, and therefore consisted entirely of written
texts where direct intertextual references should reasonably be expected.
There was only exception to this relationship between the ontogenetic
sequence of development and the type of intertextual references in ALABS.
The Class Loans/Returns consisted of a group of workpractices relating to
Tutors that were actually transplanted directly from the Manual System into
ALABS. This series of workpractices were the most 'manual' group of
workpractices in ALABS and might help explain its reliance on dialogic
intertextuality .
6. CONCLUSIONS
In order to describe the relationship between an information system and

its organisational context, a systemic semiotic approach has been described.
Using this kind of organisational semiotics to characterise an information
system requires the description of its workpracticesin terms of its multiple
constituent genres. New concepts were developed to enable the description
of information systems features in terms of multiple genres. Explicit
between-genre linkages joining several sets of genres associated with
ALABS system features were introduced. Genres involved in these
arrangements are referred to as genre assemblages, and the links between
them are called genre associations. Genre associations are persistent links
between genres that are ideologically motivated. A genre association may
not necessarily be apparent during the normal conduct of a workpractice, but
may become so when the workpractice is being renegotiated. Having defined
the relevant genres, genre assemblages, and associations for the
workpractices associated with a system in its organisational contexts, a
composite view of the system called a genre collection can be produced.
In accounting for the large-scale structure of workpractices associated
with information systems, three kinds of intertextuality were identified:
direct intertextual relations, elliptical intertextual references, and dialogic
intertextual relations. Each played a significant role in linking discrete
workpractices into larger assemblages. Assemblages and associations have
proven to be significant in explaining the diachronic or ontogenetic
development of systems. A number of principles emerged as a consequence
of tracing the development of ALABS workpractice assemblages. These
principles now need to be applied to other systems to see if they represent
fundamental aspects of higher order structure in workplaces or simply
tendencies within the diachronic development of the system considered here.
Systemic Semiotics provides an explicit contextual theory and a related set
of methods that helps to explain how social subjects, in negotiating various
systems features, enter into an already existing network of social actions and
activities, social relations of power, and textual strategies.
ACKNOWLEDGMENTS
The author gratefully acknowledges the assistance of Dr. L. Schafe in

implementing the Virtual Reality of the Microcomputer Laboratories used in
Figure 2. I am indebted to my colleagues at the Microcomputer Laboratories
who have been the subjects of - and subjected to - my analyses for many
years.
82 Rodney J Clarke
BIBLIOGRAPHY
Althusser, L. 1971, Lenin and Philosophy and Other Essays, London: New Left Books.
Clarke, R. J. 1995, (Hyper)text in Context: Applying Systemic Functional Linguistics to
Hypersystems Development, Program, Abstracts of Invited Speeches and Position
Statements- Workshop on Social Contexts of Hypermedia Department of Informatics,
Umea University, Sweden, 16th-17th February, 1995.
Clarke, R. J. 1996, The Persistence of Systems in Organisations: Genre Analysis of Systems
Decommissioning, in Holmqvist, B.; Andersen, P. B.; Klein, H. and R. Posner eds/ 1996,
Signs of Work: Semiosis and Information Processing in Organisations Foundations of
Communication and Cognition Grundlagen Der Kommunikation und Kognition, Series,
59-91.
Clarke, R. J. 1998, Systemic Semiotic Development Using SFX: Prototyping Genre-Based
Representations ofiS Workpractices, ISIC/CIRAiISAS'98: IEEE International
Symposium on Intelligent Control ISIC" held jointly with International Symposium on
Computational Intelligence in Robotics and Automation CIRA, and Intelligent Systems
and Semiotics ISAS" National Institute of Standards and Technology, Gaithersburg,
Maryland, USA, September 14-17,1998, pp.846-851.
Clarke, R. J. 1999, Systemic Semiotic relations between Workpractice Actions and Texts,
Sign Processes in Complex Systems: 7th International Congress of the IASS-AIS:
International Association for Semiotic Studies/ Association Internationale de Semiotique
Dresden, University of Technology, October 6-11, 1999 forthcoming.
Clarke, R. J. 2000, An Information System in its Organisational Contexts: A Systemic
Semiotic Longitudinal Case Study, PhD Dissertation, Department of Information Systems,
University of Wollongong, Australia.
Clarke, R. J. and L. Athanasiadis 1987, Automated Library and Borrowing System ALABS,:
Implications for CAL, in Barrett, 1. and J. Hedburg eds/ 1987, Using Computers
Intelligently in Tertiary Education: Proc. Conf. CALITE-87 ASCILITE, November 29th-
December 3rd.
Clarke, R. J.; Athanasiadis, L. and C. Nicastri 1989, ALABS Version 3: Procedures and Users
Manual Microcomputer Laboratories, University ofWollongong, 242 pp.
Crystal, D. 1985, A Dictionary of Linguistics and Phonetics, 2nd Edition Oxford, UK: Basil
Blackwell.
de Certeau, M. 1984, The Practice of Everyday Life, Translated by Steven Rendall, Berkely:
University of California Press.
Fairclough, N. 1992, Discourse and Social Change, Cambridge, UK: Polity Press and Basil
Blackwell.
Fawcett, R. 1986, Forward in Ventola, E. 1987" The Structure of Social Interaction: A
Systemic Approach to the Semiotics of Service Encounters Open Linguistics Series,
London: Frances Pinter.
Foucault, 1972, The Archaeology of Knowledge, Translated by A. M. Shendon Smith,
London: Tavistock Publications.
Freadman, A. and A. Macdonald 1992, What is this thing called Genre? Four Essays in the
Semiotics of Genre, Mount Nebo, Queensland Australia: Boombana Publications.
Halliday, M. A. K. 1978, Language as Social Semiotic: The social interpretation of language
and meaning, Great Britain: Edward Arnold.
Halliday, M. A. K. 1994, Introduction to Functional Grammar, 2nd Edition, Edward Arnold.
Halliday, M. A. K. and R. Hasan 1976, Cohesion in English, London: Longman.
Halliday, M. A. K. and R. Hasan 1985, Language, context and text: Aspects oflanguage in a
social-semiotic perspective, ECS805: Specialised curriculum: language and learning
Victoria, Australia: Deakin University.
Halliday, M. A. K. and J. R. Martin 1993, Writing Science: Literacy and Discursive Power,
Critical Perspectives on Literacy and Education, London and Washington: The Falmer
Press.
Hodge, R. and G. Kress 1988, Social Semiotics, Polity Press.
Holmqvist, B. and P. B. Andersen 1987, Work Language and information technology, J.
Pragmatics 11,327-357.
Kling, R. and W. Scacchi 1982, The Web of Computing: Computing Technology as Social
Organisation, Advances in Computers 21, New York: Academic Press.
Kress, G. 1985, Linguistic processes in sociocultural practice, ECS806 Sociocultural aspects
oflanguage and education, Victoria: Deakin University.
Kress, G. 1988, Language as Social Practice, in Kress, G. ed/ 1988, Communication and
Culture: An Introduction UNSW Press, 79-129.
Lemke, J. L. 1995, Intertextuality and Text Semantics, in Fries, P. H. and M. Gregory eds/
1995, Discourse in Society: Systemic Functional Perspectives Meaning and Choice in
Language: Studies for Michael Halliday Norwood: Ablex Publishing.
MacWhinney, B. 1995, The CHILDES Project: Tools for Analyzing Talk, Second Edition
Hillsdale, New Jersey, USA: Lawrence Erlbaum Associates, Inc., Publishers.
Martin, 1. R. 1992, English Text: System and Structure, Amsterdam: John Benjamins.
Martin, 1. R. 1994, Genre, ASFL Spring School 1994- Level III Genre, Brisbane 28th
September: Day 3, Session 9, 20pp.
Schirato, T. and S. Yell 1996, Communication and Cultural Literacy: An Introduction,
Australian Cultural Studies Series, St. Leonards, Australia: Allen and Unwin.
Thibault, P. 1. 1991, Social Semiotics as Praxis: Text, Social Meaning Making, and
Nabakov's Ada, Theory and History of Literature, Volume 74 Minneapolis: University of
Minnesota Press.
Todorov, T. 1984, Mikhail Bakhtin: The Dialogic Principle, Translated by Wlad Godzich
Theory and History of Literature, Volume 13, Minneapolis: University of Minnesota
Press.
Ventola, E. 1987, The Structure of Social Interaction: A Systemic Approach to the Semiotics
of Service Encounters. London: Frances Pinter Publishers.
Chapter 4
Actability: a Way to Understand Information Systems

Pragmatics
Goran Goldkuhl and Par J. Agerfalk
1. INTRODUCTION
The many failures of IT-based information systems give rise to a strong

imperative for researchers in information systems and neighbour disciplines
to create a better understanding of the nature of such systems and their
organizational use. Just viewing an information system (IS) as a technical
black box having some social and organizational effects is not enough. We
must understand information systems in a deeper sense than just as one kind
of technical artefact. What special kind of artefact is an information system?
The semiotic perspective offers great possibilities for such a deeper
understanding (Stamper, 1994; 2001). Starting from the notion of sign, we
can focus on different aspects of information systems as sign systems,
ranging from pure technical to social and organizational issues. In classical
semiotics, three levels (or aspects) are distinguished: syntactics, semantics
and pragmatics. Stamper (ibid.) adds some more levels and constructs a
'semiotic ladder', with the following steps from bottom to top: physical
world, empirics, syntactics, semantics, pragmatics and social world.
The pragmatic dimension of human communication has been studied and
conceptualized within speech act theory (Austin, 1962; Searle, 1969;
Habermas, 1984). A 'language action' perspective on information systems
has been articulated by several scholars, taking their main inspiration from
86 Goran Goldkuhl and Par J Agerfalk
speech act theory (e.g., Winograd & Flores, 1986; Goldkuhl & Lyytinen,
1982; Dietz, 1994). The fundamental speech act thesis is that communication
is to be seen as one kind of action. Searle (1969) distinguishes between four
different sub-acts of a speech act:
Utterance act
Propositional act
IIIocutionary act
Perlocutionary act
The utterance act is to be seen as the production of a sequence of words
that together form a comprehensible wholeness of an utterance. We
understand this mainly as equivalent to the syntactic level. The propositional
act means referring and predicating, i.e., representing a world talked about in
an utterance. This corresponds to the semantic level. The iIIocutionary act is
what we are doing by speaking, for example, stating, commanding,
promising or declaring. The perlocutionary act is the intentional 'causing' of
effects in listeners. The relationships between these two last sub-acts and the
semiotic aspects are not straightforward. We will discuss these matters in
Section 2 below.
To distinguish between utterance, proposition, iIIocution and perlocution
seems to be important. We think, however, it is misleading to describe these
as different sub-acts performed within a speech act. They are rather different
aspects of a speech act. We introduce these notions here as we will use them
later in the chapter.
A pragmatic perspective on information systems seems to offer good
possibilities for a deeper understanding of such systems as artefacts in
organizational settings. The purpose of this chapter is to explore the
pragmatic character of information systems. Taking the semiotic ladder as
our point of departure we eventually arrive at the concept of information
systems actability, which we define as an information system's ability to
perform actions, and to permit, promote and facilitate the performance of
actions by users, both through the system and based on information from the
system, in some business context. In conducting the journey, we investigate
possible meanings of pragmatics related to information systems and touch
upon some implications for information systems design.
The chapter is structured as follows. In Section 2 we discuss different
interpretations of information systems pragmatics starting from the semiotic
ladder. In Section 3 we relate information systems to concepts of action,
organization and artefact. A generic action model is presented as a basis for
the actability concept. The key concept of this chapter, information systems
actability, is thoroughly discussed in Section 4 and some implications for
information systems design are presented in Section 5. Finally, conclusions
of the chapter are presented in Section 6.
Actability: a Way to Understand Information Systems Pragmatics 87
2. INFORMATION SYSTEMS AND PRAGMATICS
Stamper (1994; 2000) presents a semiotic ladder for analysing different

aspects of information systems as sign systems. This Organizational
Semiotics approach to information systems has had a substantial influence
on the IS field and several authors have adhered, and contributed, to it (cf.
Liu et aI., 2000). The semiotic ladder consists of the following steps:
physical world, empirics, syntactics, semantics, pragmatics and social world.
The pragmatic level is defined in the following way (ibid.): 'Pragmatics is
the branch of semiotics concerned with the relationships between signs (as
meaningful utterances) and the behaviour of responsible agents, in a social
context.'
The relationship between action (i.e., 'behaviour of responsible agents')
and signs is, however, not unambiguous. It is possible to identify different
relationships. We use Figure 1 to illustrate our reasoning.
A message (depicted in Figure 1) can be produced by a human agent or
an IS. This message (a sign constellation) is received and interpreted by
some human agent. After interpreting the message, the agent performs some
action based on that interpretation. The common understanding of the
pragmatic aspect of information systems seems to be this kind of 'ex post
action', i.e., an action based on messages from an IS (e.g., Langefors, 1995).
The pragmatic aspect of information systems is concerned with the actions
performed by humans after receiving messages from information systems. In
what ways are humans influenced by messages from information systems?
What are humans doing when using messages from information systems as a
base? This is one interpretation of pragmatics, but not the only one.
Let us look more closely at the message, which can be produced by an IS
or a human. Is such a message not the result of an action? Yes, it should be
viewed as an action result. We must, thus, add another pragmatic dimension
to the one described above: the message as a result of an action. The
production oftl1is message is to be considered as an action. What else would
it be? This is obvious if the producer of the message is a human agent. In this
case, the situation is a human-to-human communication. Speaking and
writing is action (Austin, 1962; Searle, 1969). Even when the producer is an
IT artefact, the production of the message must be a communicative action.
This might perhaps be harder to accept, but we will elaborate on this below
(Section 3). This thesis is vindicated in the language action tradition of
information systems (e.g., Goldkuhl & Lyytinen, 1982).
88 G6ran Goldkuhl and Par J. Agerjalk
Message
(Sign)
Action within the Action based on the

sign transfer transferred sign
Figure 1. Different kinds of actions related to signs
The sign transfer (i.e., a communication process) is an action, according

to our definition. The action within the sign transfer should be seen as a
pragmatic aspect. This means that the information system's production of
messages to its environment should also be seen as an action. Pragmatics is
not only related to actions based on messages from information systems. The
very production of messages (signs) also has a pragmatic dimension, since
this sign production should be considered as action.
The sign transfer as action is, with reference to concepts of speech act
theory (Austin, 1962; Searle, 1969), considered to be a so-called
illocutionary act. This is what is done in speaking. Actions based on the
communicated sign are considered to be on the perlocutionary level. In
speech act theory, this relates to effects arising from communication: what
the listener does covertly or overtly.
These two aspects are important pragmatic aspects of information
systems:
action within the sign transfer
- action based on the transferred sign
These two aspects should be both acknowledged and distinguished, as
they can be by means of the speech act notions of illocution and perlocution.
As can be seen from Figure 1, they are related. We will elaborate this
further. The sign as action might give rise to expectations in the interpreter.
These expectations are intrinsically tied to the illocution performed. The sign
is a base for action and this action can be seen as fulfilment or rejection of
the expectations as well as other perlocutionary effects. When a person A
asks a person B 'Would you bring me a glass of water?', A performs a
request directed towards B. This request action (which is a special kind of
illocution) gives rise to expectations directed from A to B. A expects B to
perform an action of bringing water. B can react in different ways, declining
the request or fetching a glass of water. Refusal or delivery is action in
response to the request. These different types of actions are considered as

perlocutionary effects of the speech act of the request. The delivery of the
water is the effect intended by A.
In our conception of information systems pragmatics, we include the two
action aspects described above. Stamper (1994) treated these two aspects
within his semiotic theory. In Stamper's semiotic ladder the illocutionary
aspect seems mostly to be related to the pragmatic leveP and the
perlocutionary aspect seems mostly to be related to the social level 4 • In our
framework it is not possible to make these kinds of characterizations. The
two types of action (action within sign transfer, action based on transferred
signs) are both actions and as such they of course involve a pragmatic
dimension. We claim that these two kinds of action involve a social
dimension also.
The sign transfer action is a social action since its main purpose is to
create a shared understanding between human beings (Habermas, 1984).
Different interpersonal relationships are established because of the actual
illocution performed. The action based on the transferred sign is also to be
seen as a social action. This is the case even if the listener does not explicitly
respond to the 'initial' speaker overtly. Even if the listener performs an
action (based on the transferred sign) in isolation, we call this a social action
since the user directs the action in relation to other humans. This seems to be
in accordance with Weber's (1978) classical definition of a social action.
Pragmatics is about action. When we talk about intentionally arranged
signs related to action, it seems impossible to exclude a social dimension.
The actions of the two kinds are social actions. In our opinion, it is hard to
distinguish between the social and pragmatic stages on the semiotic ladder
concerning these types of actions. Producing signs requires actions and such
actions are in most cases social actions.
3 Stamper (1994) explicitly relates the pragmatic level to the illocutionary level: 'It is also
possible to model some aspects of intentional sign use in terms of speech act theory by
studying so-called illocutionary verbs'.
4 When describing the social level, Stamper (1994) makes an explicit reference to the
perlocutionary level: 'Each illocutionary act will have a social consequence achieved by
the listener(s) performing (a) perlocutionary act(s), which change(s) the social world.'
(ibid.).
90 Goran Goldkuhl and Par J. Agerjalk
3. INFORMATION SYSTEMS: ACTION

ARTEFACTS IN ORGANIZATIONS
3.1 Human action
Human action is about making a difference. This is one fundamental

thesis of the American pragmatists (Thayer, 1981). Humans intervene in
their outside world and create differences. Human action often aims at
making material changes. Chopping wood means that wood is transformed
into firewood, which later can be put into a fire to create heat. Action is
meaningful and intentional. Humans act in order to achieve ends.
The world of humans is not only a material world. It is also a social
world consisting of other humans and their expressions. Communicating is
also action (Austin, 1962; Searle, 1969). To utter a sentence, directed
towards another human being, is one kind of action. There are differences
and similarities between communicative and material acts. It is, however,
possible to use a generic action modeJS to describe both kinds of action
(Figure 2).
We distinguish between the following categories:
- Action prerequisites (external and internal)
- Actor (interventionist)
- Acting (performance of action)
- Action result (what is done)
- Receiving (interpreting) action result
- Actor (recipient)
- Effects of action
- Relationships between actors
Action context (place and time)
We use two examples to describe the generic model. (1) A
communicative act of a woman asking her husband to bring in some
firewood: 'Would you please bring some firewood in. I think it is cold in
here'. (2) A material act of chopping wood performed by the husband.
We use the different action categories described above and in Figure 2 to
characterize the two kinds of actions (see Table 1 for an overview). This
does not mean that there are not important and fundamental differences
between communicative and material actions. There are important
differences, which will be shown. In communication we transform our own
knowledge (as the base) into a linguistic utterance (the result). When doing
so we are using language as an instrument. Language must be inter-
5 This generic action model emanates from earlier work (Goldkuhl, 1998ab; Goldkuhl &
Rostlinger, 1999).
subjective to a large degree, otherwise the communication will not be

successful, i.e., no shared understanding will arise. In material action an
external base (such as wood) is transformed into a material result (such as
chopped firewood). The actor is using an external instrument (such as an
axe) to produce results. The utilization of such an instrument (tool) extends
the capabilities of the actor. The actor will have different relations towards
the different external objects. He or she will have a transformative relation
towards the base, a utilizing relation towards the instrument and a producing
relation towards the result. Base, instrument and result are thus important
categories for describing action. The base is transformed to a result by a
competent utilization of an instrument. In the example, there is an oral face-
to-face communicative action; and in such a situation there is no use of an
external instrument. When speaking to a person at some distance we usually
require some instrument, often a telephone, to bridge the distance. Written
communication implies the use of external instruments, such as a pen or
typing equipment.
Action relationship
Identity
"
.. .. ..,,----- ........ ..
...
...
Identity
Values, norms Values, norms
k"
rAl
& preferences ~ & preferences
Abilities
'""'J'"' Abilities
11I.
Emotions Emotions
t )'--__
Deliberations, Deliberations,
intentions & plans A intentions & plans
"'"' .J
Situational Situational
~e
comprehension comprehension
& attention Actor Actor & attention
Conditioning
& reflexivity
Action context
Figure 2. A generic model of social action
Base and instrument are two fundamental prerequisites for action. In the
case of material action, these prerequisites are external objects. In the case of
communicative action, these prerequisites are internal elements, i.e., parts of
the practical consciousness of the actor. There are other internal
prerequisites, both situational and trans-situational. We claim that actions are
purposive, i.e., there are intentions associated with the actions. We do not,
however, claim that intentions are always well defined and deliberate before
the performance of the action. Intentions can arise from situations (Joas,
92 G6ran Goldkuhl and Par J. Agerjalk
1993). Ends and means can be figured out when interacting with the
environment. This implies that the situational understanding and the
attention of the actor are fundamental for acting (ibid.; Giddens, 1984).
Table 1. A comparison between a communicative act and a material act (example)
I Category I CommuDlcallve actIon (request) I Matenal acllon (choppmg)

ITnterventlOntst actor Jl1e WlTe lbe husband
i,7ntentlOn To urge the husband to fetch firewood To fullll the wife s request
IActmg I Speakmg (request for fjrewood) ! Choppmg
I Base I Knowledge about needs Wood

i,7nstrument [Language AXe
'AliilItles I Commumcallve competence Choppmg competence
SituatIOnal comprehenSIOn & I Percelvmg room temperature and non- Contmuously momtonng own
attention existence of firewood chopping
Result of actIon Utterance (request) I Forewood (chopped)
ReCIpIent The husband lhe wlte
iEJJects JI1enusoano uooerstanas aoo accepts [m The wife receIves tlrewood to put
the wife's request and goes to chop on the fire
wood (2) Heat from the fire
Following the classical model of social action by Weber (1978), we can

add that actions are also governed by values, emotions and traditions
(norms). In accordance with the writing of Weber (ibid.), we claim the
multifunctionality of action (Goldkuhl, 1998a); i.e., action is often governed
by all these prerequisites (purposes, values, emotions and norms), which are
expressed during action. The request of the wife is of course purposive, but
at the same time it can be seen as compliance to a social norm concerning
division of work within this family, namely 'the husband is the one who
fetches firewood'. This is also related to the conceived identity of the actors:
'Who am I-what is my role in this situation?' The action prerequisites
mentioned by Weber (ibid.) are important, however they are not complete.
As can be seen in Figure 2 we add some more important prerequisites.
The two actions described together form a social interaction. Of course,
we interact with other people through language, but we also interact by
producing and delivering material objects to each other. The material action
of the husband is assigned by the request of the wife. The wife's request is
self-assigned. In the interaction, different action relationships arise between
the participants6 • The request gives rise to expectations directed towards the
husband. The delivered firewood may give rise to emotions of gratitude
towards the husband.
There is an important difference in the generic action model between
action result and action effects. The action result (what is done) is within the
range of the actor, i.e., it is within the actor's control. The effects are what
6 Habermas (1984) has contributed to speech act theory by emphasizing that different
interpersonal relationships are established between speaker and hearer in communication.
can arise as consequences of the action. The speaker has no control over the
effects connected to the receiver. The listener can understand the utterance in
different ways; e.g., they can misunderstand in ways that the speaker could
not presume. The listener can act in ways outside the scope of the speaker's
intention. In the communicative act case, the action result reflects the uttered
request. The effects of the action are the understanding of the receiver and
the consequential actions based on this understanding 7 • The result of the
material act is the chopped firewood. The effects are what the wife does
when she receives the firewood. The reception of a message (oral or written)
always implies an interpretation of what is communicated. Even the
reception of material objects implies interpretation: 'What kind of material is
this and how can I use it?' Effects of actions may be both social/non-material
and material. This applies to both communicative and material actions. The
communicative action of the wife-the request-gives rise to both
commitments made by the husband and material changes of chopped wood.
The material action of delivering firewood to the wife gives rise to social
effects (the fulfilment of a performed commitment and the gratitude for this)
and material effects (heat from the fire).
This generic model of action can be compared to other models. It is
beyond the scope of this chapter to make any thorough comparison to other
models 8 • We will, however, briefly compare it to a model by Norman (1988).
We do this since we will discuss the actability concept in relation to
Norman's model in Section 4 below. Norman (1988) presents a seven-stage
model of human action. The main domain for his action model is humans'
interaction with everyday things. The seven stages are (1) forming the goal,
(2) forming the intention, (3) specifying an action (action sequences), (4)
executing the action, (5) perceiving the state of the world, (6) interpreting
the state of the world and (7) evaluating the outcome. These seven stages can
be divided into three phases: preparation (stages 1-3), execution (stage 4)
and evaluation (stages 5-7). Norman speaks only of execution and
evaluation, but we think it is also important to distinguish between
preparation and execution.
There are some terminological differences between the models. Norman
differentiates between the goal and the intention. The differences between
these seemingly similar concepts are that goals are to be seen as the ends,
7 These differences can be related to concepts of speech act theory. Searle (1969)
distinguishes sharply between what is done within the utterance (the illocution) and what
is intended to be done by the listener (the perlocution). The illocution is one part of the
utterance, i.e., what is done by the speaker (the result). Effects related to the listener can be
named perlocutionary effects.
8 Parts of the theoretical background of this action model are described by Goldkuhl (1998a;
1998b) and Goldkuhl & Rostlinger (1999).
94 Goran Goldkuhl and Par J. AgerJalk
and intentions are to be seen as the means. In our action model we

distinguish between underlying values and preferences (i.e., more
fundamental endeavours and purposes), specific intentions of the action
(ends), plans (means for achieving the intentions) and deliberations (which
involve considerations concerning both ends and means, i.e., reasons for
choosing a certain action and not choosing another one).
Norman's model can be criticized for having a rationalistic flavour: a
strict sequence of preparation, execution and evaluation with predefined
purposes before entering into acting. In our model we emphasize the
continuous perception of the world, i.e., the world is perceived before,
during and after the execution of action. An actor starts from a situational
understanding of the action context. This situational understanding
determines what ends and means will be chosen9 • (The wife is thinking 'It is
cold in here, but I could make a fire in the fireplace'). During action, the
actor is continuously monitoring his or her actions, the results and effects.
This constant monitoring makes it sometimes possible to modify an action
during its execution. The actor is attentive in action. When chopping wood it
is necessary to be attentive otherwise one can cut oneself. This notion of
continuos perception is also acknowledged by Norman (ibid.) who states that
'There is a continual feedback loop, in which the results of one activity are
used to direct further ones, in which goals lead to subgoals, intentions lead to
subintentions. There are activities in which goals are forgotten, discarded, or
reformulated. '
We have above described some fundamentals of communicative and
material action. Not all human action is of these types, which can be labelled
interventionist action (Goldkuhl, 1998b). There are other types of action,
such as investigative action and reflective action. These kinds of actions do
not aim at external changes, but rather at internal changes of the actor in
terms of improved knowledge (ibid.).
3.2 Organizational action
Organizations create products (goods or services) beneficial for their

environment; i.e., for their customers or clients. In doing this, different
actions must be performed. There can be material acts with the purpose of
producing material results. There can be communicative acts with the
purpose of producing communicative results (utterances and messages).
Material acts and communicative acts together form patterns of action. They
9 We think it is possible to combine a purposive view of action with a situational view, i.e.,
that action is both purposive and situationally constrained. Purposes can arise from
situations, but can also be brought into situations by deliberate planning.
fonn not only a language game, but also an 'activity game', with relations
between communication and material treatment (Goldkuhl, 1996). Large
parts of this activity game are recurrent actions and are thus institutionalized
in the organization and the practical consciousness of its participants (Berger
& Luckmann, 1967; Giddens, 1984; Scott, 1995). Some of these actions are
internal, i.e., directed towards the interior of the organization. Other acts are
external, i.e., directed towards its environment.
When we use the tenn organizational action, what do we mean by this?
Do we mean that the organization acts? Or do we mean that the human
members of the organization act? Actually, we use both meanings. We
conceive organizations to have action ability (Goldkuhl & Nilsson, 2000).
We say that organizations act. Is this not a reified lO position? A view that
organizations have been given an ontological status of their own outside the
realm of human originators? We do claim that organizations are actors, but
they are not actors on their own. They always have a human origin and
purpose. Organizations are created by humans and for the purposes of those
humans. Organizations act always through their human co-workers or
through artefacts arranged by humans. They cannot act by themselves, only
through humans (ibid.; Ahrne, 1994).
The co-workers act on behalf of the organization. They act in the name of
the organization. An act perfonned by a human co-worker is always dual. It
is an action perfonned by a human being, but is also at the same time an
action perfonned by the organization. Humans act in organizational roles.
Their action is representative. They represent the organization when acting.
Given this definition, it is possible to say at the same time that an
organization acts and that its human co-workers act.
3.3 The nature of artefacts
When describing human action above, we pointed out that instruments

are often important means for perfonning actions. To chop wood without an
axe is not possible. Humans use artefacts to extend their action ability. Some
artefacts improve actions and others even enable certain actions. The modem
world is a world crowded with artefacts. They play a dominant role in
today's society (Latour, 1992). Many artefacts go even beyond enabling
human action. They are created to be independent perfonners of action. Such
artefacts have dynamic properties, which make it possible for them to
function on their own. A modem washing machine can work by itself after a
10 Berger & Luckmann (1967) describe the meaning of reification and the problems and
dangers of reifying social phenomena, i.e., disregarding the human origin of socially
constructed products.
human has started it. The washing machine substitutes the washing work
otherwise performed by a human being. The axe does not make anything by
itself. It is a tool with static properties. Weizenbaum (1976) uses the
concepts of prosthetic tools and automatic machines. A prosthetic tool, like
an axe, extends the ability of humans. An automatic machine, like a washing
machine, has an autonomous ability to function on its own. There are,
however, artefacts that do not easily fit into these categories. How about a
car? It has abilities to perform work (like moving), but it does not function
totally on its own. A human drives a car. A car requires constant
manoeuvring. It is a dynamic tool, but a human must operate it.
We divide artefacts into these three categories (Table 2). We will use this
classification when discussing information systems based on computers and
information technology.
Table 2. Classification of different types of artefacts
Static tool Mechanical tool Automatic machine

Way offunctioning Static Dynamic Dynamic
Degree of dependence Active use by human Active use and Autonomous
surveillance by human
Type of action Human-supported action Human-artefact Artefact action
cooperative action
Role of humans Wield Manoeuvre Initiate
Example Axe Car Washingmachine
3.4 IT-based information systems
IT -based information systems can be used for communicative action or

for internal or external organizational ac.tions. The IS actions must be well
integrated into the activity games of the organization. This means that an IS,
as an artefact, must be congruent to the actions of humans and to the overall
objectives of the organization.
We identify three kinds of actions related to information systems:
interactive, automatic and consequential actions (Figure 3). A user can
utilize an information system to perform certain actions. The IS is used as an
action medium for the user to perform actions. In interaction between the
user and the computer a communicative action is performed and the results
of this action (i.e., a message) is transmitted through the IS. This kind of
action-which we call an interactive action-is performed with support by
the IS and through the IS.
An IT -based information system is not only used as an interactive
mechanical tool, but also as an automatic machine. Usually, a system is
provided with an ability to perform certain predefined actions of its own.
Based on messages in the system, new messages can be derived and
communicated to its environment. These kinds of action are called automatic

actions. It is important to state that the computer performs these actions in
accordance with well-defined rules created by humans. The responsibility
for the actions rests upon those humans who decide the rules to be followed.
We believe that it is possible to talk about automatic actions performed by an
IT -based information system in this way, without falling into the trap of
reification. In this context, reification would mean that we forget the human
origin of information systems and IS action, that we conceive information
systems to be independent actors creating actions totally on their own. We
avoid such a reifying position by claiming that information systems perform
actions that are predefined to them by human actors. An IS does not find out
what kind of actions to perform. The systems' actions derive ultimately from
the rules predefined to the system. We do not presuppose computers to have
human properties of consciousness and ethical responsibility.
IS ~
Interactive action Automatic action Consequential action
Figure 3. Different actions related to an information system
Automatic actions of an IS will produce messages (with defined

illocutionary and propositional meanings) to its environment, i.e., to some IS
users. Such messages can be a basis for actions by these users. We call such
user actions consequential actions. These actions can be seen as
perlocutionary effects in relation to the communicative actions performed by
the IS.
This view of information systems means a rejection of the view of
information systems as mere 'containers of facts', as passive providers of
information for future actions of human users. Information systems are used
for communication among people within and outside the organization. An
information system is also an active party in such a communication. It is an
artefact with capabilities to perform communicative actions on behalf of the
organization (Goldkuhl & Agerfalk, 1998). We conceive information
systems to be information action systems. An IS should have its own
pragmatic properties (i.e., be able to perform communicative actions with
illocutionary intent) and not only bring about pragmatic effects (i.e., humans
performing actions based on messages from information systems as
'perlocutionary responses'). This VIew IS also reflected in our

conceptualization of 'message'.
In traditional information systems theory (e.g., Langefors, 1966), the
smallest element of information is the elementary message (the e-message).
The basic e-message is a triple (0, p, t) consisting of a reference 0 to an
object in the universe of discourse, a reference p to a property predicated to
that object and a reference t to a point in time or a time interval during which
the e-message is valid. Similarly, a relational e-message e = ((o\, 02, ... , On),
r, t) is used to assign the objects (0\, 02, ... , On) to the relation r. Goldkuhl
(1995) argued that the e-message concept must be extended with an
illocutionary component to reflect the illocution of the message. By
elaborating this further, we arrive at the concept of action elementary
message (ae-message). When we talk about messages as results of
communicative actions, we refer to ae-messages. Since an illocution can be
(and usually is) associated with a larger propositional content (references
and predicates) than a single e-message, an ae-message can be considered as
an aggregate of e-messages with an associated action mode representing the
illocutionary intention of the ae-message. The key point is that while an e-
message is the smallest unit that carries information (or rather propositional
content), an ae-message is the smallest unit that carries an action mode.
When discussing the pragmatics of information usage in organization, it is
meaningless to dissect information down to the level imposed by the e-
message concept. However, when it comes to the design of the persistent
storage of information (e.g., a database) it is important to understand the
relation between ae-messages and e-messages so that 'traditional'
approaches can be utilized. It is important, though, to not fall into the trap of
disrespecting the coupling between the propositional content and the action
mode.
Let us consider an example to illustrate the concept of ae-message.
Suppose that a virtual bookshop is, via their Web site, offering potential
customers a special 'book of the month' to buy for a very special price. This
on-line offer is an ae-message sent from the bookshop to potential
customers. The action mode of the message is a promise that customers can
buy the book at the specified price during the specified month. The
propositional content might be data on the book's title, number of pages,
author, publisher, price (ordinary and reduced), et cetera. Note that we would
require several e-messages to represent this single ae-message' s
propositional content and that it would not have made sense to try to connect
these separately to its action mode. Doing so would have required another
relational e-message referencing the first ones. This would have been to treat
illocutions as describable pieces of reality (belonging to the object system)
rather than as constituents of communicative actions (as parts of the

discourse).
4. THE CONCEPT OF INFORMATION SYSTEMS

ACTABILITY
Information systems actability (Goldkuhl & Agerfalk, 1998; Cronholm et

aI., 1999; Agerfalk, 1999) is a conceptualization of the understanding of
information systems as tools for organizational action and communication.
This is in contrast to the misconception that information systems are used
only as passive providers of information to be used for future action, i.e.,
action based on transferred signs. Actability is an attempt at a synthesis of
the usability perspective as applied within the human-computer interaction
tradition (e.g., Preece et aI., 1994) and the language action perspective
(LAP) (e.g., Goldkuhl & Lyytinen, 1982, Winograd & Flores, 1986) from an
information systems point-of-view.
A central concept for actability is the performance of communicative
actions, referred to as elementary actions (e-actions). The result of an e-
action is an ae-message. As discussed above (Section 3.4), there are three
kinds of such actions related to information systems: interactive, automatic
and consequential.
Let us again consider the bookshop example introduced above. A user
accessing the Web site and reading about the 'book of the month' would be
performing an interactive action by purchasing the book, utilizing an order
function provided by the Web-based information system. If the user, on the
other hand, chose to print an order form and send it by traditional post, this
would be a consequential action. Suppose also that the bookshop' s
information system collected all orders at the end of the day and sent a
supplier order via Electronic Document Interchange (EDI) to a publisher
containing all books ordered that day from that publisher; it would have
performed an automatic action.
Based on this discussion, 'Actability' can be defined as: an information
system's ability to perform actions, and to permit, promote and facilitate the
performance of actions by users, both through the system and based on
information from the system, in some business context. (cf. Cronholm et aI.,
1999).
100 Goran Goldkuhl and Par J. Agerfalk
4.1 Information system usage situations
When talking about the performance of actions in relation to information

systems, and thus usage of the system, it is possible to distinguish three
different types of IS usage situations:
1. An automatic usage situation occurs when an information system's
'ability to perform actions' is considered. That is, when an IS performs e-
actions without any direct human intervention, but according to
instructions by, and on commission of, some human actor. In these
situations the IS is used as an automatic machine.
2. An interactive usage situation occurs when an IS is used to 'permit,
promote and facilitate the performance of actions by users ... through the
system'. In these situations the IS is used as a mechanical tool.
3. A consequential usage situation occurs when an IS is used to 'permit,
promote and facilitate the performance of actions by users ... based on
information from the system'. In these situations the IS is used as a static
tool.
In all three types of IS usage situations, there are three basic components
involved: (1) the information system used by (2) an actor to perform (3) e-
actions. These three entities participate together in a ternary relationship that
forms the usage situation. To understand these components and how they
interact within the ternary relation is the key to understanding actability.
To achieve such understanding it is necessary to comprehend the first
component, information systems, as systems for communicative actions-
that is, as information action systems. Thus, we define an 'information
system' as: a technically implemented social system consisting of an action
potential (a repertoire of actions and a vocabulary), a memory of earlier
actions and action prerequisites, and actions performed interactively by the
user and the system and/or automatically by the system. By using
'technically implemented social system' we adhere to the argument of
Goldkuhl & Lyytinen (1982) that information systems are not technological
systems, but 'social systems only technically implemented'. To say that
information systems have an 'action potential' means that any information
system per se defines the set of actions that can be performed by using it.
This also incorporates the vocabulary used to express communicative actions
within it-how to refer to the state of affairs that is talked about. By
'memory of earlier actions and action prerequisites' we refer to the persistent
storage aspect of information systems, which is updated and extended
through the performance of e-actions. This is an important feature of
information systems used to keep track of what has been said and done
through the system and serves as important background for future actions.
One can think of this action memory as a repository of messages held by the
system. The action memory would typically be implemented as one or

several databases, even though it might make use of whatever technological
solution is available. In the bookshop example (see above), the action
memory would typically be used to store, for example, order messages from
customers in an order database. That would be an example of 'earlier
messages'. It might, however, also contain information about what book that
will be the book of the month during months to come, which then would
constitute action prerequisites. The action memories of an organization's
information systems form an important part of the overall organizational
memory. It can be used to inform about who has done what with whom,
when, and why. The 'actions performed ... ' in the definition should be
understood as the instance level of the system, whereas the 'action potential'
represents the system's type level. It is important to include both levels in
the definition since neither of them can be understood in isolation from the
other. We use information systems to do things and to talk about the world,
within the world. Information systems are not abstract models representing
the state of the world, but parts of the world themselves.
Table 3 shows how different parts of an information system belong to
these two levels. Note that the instance level consists of two different
aspects. Messages represent the state of the system while actions represent a
dynamic aspect.
The second component of an IS usage situation, the actor, relates to three
different meta-roles: the communicator, the performer and the interpreter.
The communicator is the human agent responsible for an e-action in relation
to an interpreter (recipient). Sometimes the actual performance of an action
is pursued by someone other than the communicator. This is most obvious in
the case of automatic usage situations when the information system performs
the action and thus plays the role of performer. However, sometimes another
human agent performs an e-action interactively on commission of the
communicator. This second human agent-acting as a mediator-then plays
the role of performer. As indicated above, a human agent, using the system
as an information provider for future actions outside the system (i.e., during
a consequential usage situation), plays the role of interpreter. Human agents
playing the latter two roles, i.e., performers and interpreters, are usually
referred to as users within, for example, the human-computer interaction
community. We would like to stress, however, that this notion of
information systems usage is too restrictive. From an actability point-of-
view, a communicator, as well as a performer or an interpreter, are users of
the information system.
Table 3. Description of an IT-based infonnation system with respect to its type and instance
levels
Type level Action potential: rules for what actions to perfonn, Rules for action memory
including what types of messages to treat (what types of messages to
save and the vocabulary used)
Instance level Input, intennediate, I Perfonnance of interactive Messages held in the action
output messages and automatic actions memory
(state part) (acti ve part) (state part)
The third part of an IS usage situation comprises the set of e-actions

performed during the usage situation. This set consists of zero or more e-
actions, possibly with associated sequence restrictions, that are logically
grouped together to form a coherent whole. An interactive usage situation
consists of one or several interactive, consequential, automatic and manual
actions. It is defined as: all interactive actions, and intermediate
consequential, automatic and manual actions, that are performed adjacent in
time by the same performers (cf. Agerfalk et aI., 1999). A consequential
usage situation consists of one or several consequential, automatic and
manual actions. It is defined as: all consequential actions, and intermediate
automatic and manual actions, that are performed adjacent in time by the
same performers, which are not part of any interactive usage situation.
Finally, the automatic usage situations of a system consist of one or several
automatic actions and are defined as: all automatic actions that are
performed adjacent in time by the same information system, that are not part
of any interactive or consequential usage situation.
It is important to realize that consequential usage situations as well as
interactive usage situations might incorporate human-computer interaction.
In the latter case, the interaction aims at sending an ae-message through the
system. In the former, the aim is to interpret one or several ae-messages in
order to perform an action outside the system. It is also important to realize
that interpreting earlier messages might also be an important part of an
interactive usage situation. Hence, the two types of usage are tightly
coupled, but are still distinguishable. Basically, from an information systems
perspective, an interactive usage situation is a consequential usage situation
with added complexity. The complexity is imposed by the functionality
required to formulate and send messages, and to interpret the effects of the
action. Since actions performed during consequential usage situations are
always established outside the system, this aspect is not considered part of
the IS, but might still be supported by it. What, then, do we mean by
'formulating messages'? The concept refers to the fact that before a
communicative action can be performed, the speaker must formulate the
sequence of words to utter (i.e., the preparation of the action). This includes
formulating the propositional content of the action and attaching it to an
understandable and appropriate action mode (i.e., the illocutionary meaning

of the action)lI. To interpret the effects of the action means to comprehend
how the action changed the state of the social world in which it was
performed (i.e., observing the perlocutionary effects, if possible). An
important aspect of actability is that when e-actions are performed through
an IS, these two cognitive tasks can be supported by the IS. A performer can,
for example, interactively use the system to pick items from a list of
merchandise in order to formulate an order message. It is worth noting that
this makes the action preparation explicit as opposed to implicit during
manual action.
4.2 Elementary interactions
The interactive formulation and sending of ae-messages (and, if possible,

the evaluation of their effects) is performed through elementary interactions
(e-interactions). An e-interaction is an action at a lower level of abstraction
than e-actions. An interacting user performs an e-interaction every time
something is done with the interface of the system. Each e-interaction
follows a recurring schema consisting of three phases: a user action, an IS
action and an interpretation act. The user action represents the user's direct
manipulation of objects on the computer screen, for example clicking a
button labelled 'send'. The IS action represents the information system's
response to the user action, for example sending whatever was to be sent.
This phase might include several operations performed by the IS and usually
ends with the system giving the user some sort of feedback. The
interpretation act represents the user's effort to understand what the system
achieved, by utilizing feedback from the system. We refer to this recurring
schema as the elementary interaction loop (see Figure 4). Note that the loop
might be recursively expanded since the interpretation act might require the
performance of yet another e-interaction.
Both the formulation of ae-messages and the execution of e-actions (i.e.,
sending of ae-messages) are performed through e-interactions. The latter
constitute a special case where the interpretation act includes the evaluation
of the e-action as a whole. It is important that a system give as much
feedback as possible about the effects of an e-action. However, because the
effects might not be directly observable, the system should at least make the
II Note that this should not be mistaken for a rationalistic view on actions (cf. Section 3.1).
Of course, people do not always consciously think through what to say in advance.
However, perhaps they should, at least if what they are saying constitutes important
business actions.
user aware that the e-action has been performed. Note that each e-interaction
is 'triggered' by the interacting user.
Figure 4. The elementary interaction IOOpl2
Before any e-interactions are performed, the IS might perform one or

several initial IS actions to 'set the scene'. The IS might, for example display
some initial information that the user can utilize when formulating a first ae-
message.
Figure 5 depicts the general structure of an interactive usage situation.
Based on information provided by some initial IS actions 13 , the user
formulates an ae-message through a series of e-interactions. Finally the ae-
message is sent through the IS (an e-action is executed), causing a business
effect. In each e-interaction the user is given feedback from the IS. In the
final e-interaction this includes information about the performed action and,
if possible, its business effect( s).
Within our conceptualization of IS usage, actions take place at three
different levels of abstraction. E-actions, e-interactions and the three phases
of e-interactions represent these levels. In addition, we introduced the
concept of initial IS actions above. Figure 6 depicts a JSP diagram showing
the structure of an interactive usage situation to illustrate how these
components relate.
From the JSP diagram we see how an interactive usage situation is
initialized by a series of zero or more initial IS actions, followed by zero or
more e-actions. For each e-action there is a 'preparation phase' during which
the e-action is formulated (cf. Section 3.1). This is done through zero or
more e-interactions, each consisting of a user action, an IS action (which
might in fact consist of several operations) and an interpretation act. When
the formulation is complete the user enters the 'execution phase' and can
12 Adopted from Agerfalk et al. (1999).

13 Note that these are optional, as there are interactive usage situations where no initial IS
actions are performed.
choose to execute the e-action by perfonning yet another e-interaction.

During this 'final' e-interaction the interpretation act includes the evaluation
of the complete e-action. Hence, it constitutes the 'evaluation phase' as well.
It is important to note that the user can choose to abandon the current e-
action provided the 'final' e-interaction has not yet been initiated. After that,
it cannot be cancelled.
Figure 5. An ae-message is fonnulated and sent by an interacting user during an interactive

usage situation 14
Let us return to the virtual bookshop example introduced in Section 3.4.

The bookshop is, via their Web site, offering a 'book of the month' for a
special price. By following a link on their home page, a new page is opened
and shows the 'book of the month offer' in the frame dedicated to show
details of books. A customer (i.e., an interacting user) can choose to click on
a button labelled 'add to shopping cart' and then click on another button to
go to the' cashier'. Now the user must provide the system with identification
infonnation. This is achieved by clicking on either the button labelled 'log
in' if already registered, or on the button labelled 'new customer'. Assuming
this is an old customer, when logged on, shipping details are shown in an
order document, which also contains infonnation about the book that was
put in the shopping cart earlier. To go through with the purchase, the user
must now click on the button labelled 'send order'. A new page is then
shown, telling the user that the order is being processed and that the book
should be delivered within four days if nothing unusual happens.
In the example, every single click on the different buttons constitutes an
e-interaction. The message being sent from the customer to the bookshop is
an order. The action mode of this order contains two different illocutions.
The first is a request to the bookshop to deliver the book in question. The
second is a promise to the bookshop that the agreed price will be paid when
14 Adopted from Agerfalk et al. (1999).

106 Goran Goldkuhl and Par J. Agerfalk
the book is received l5 • The order ae-message is sent by the 'final' e-

interaction performed by clicking on the 'send order' button. For every e-
interaction, the system responded by displaying a different page-helping
the user to interpret the actions. Note that the formulation of the order
message was initiated as soon as the user added the book to the shopping
cart, as the system used this information to give an order proposal. When the
user first entered the bookshop site, the 'book of the month' offer was
shown. This is an example of an initial IS action. Note that the user could
have withdrawn the purchase free of obligations, until the moment the 'send
order' button was clicked. After that the user will (probably) be legally
bound to pay for the ordered book.
!~2_
j.
Execute
the e-actlOn
0 I Cancel
.
°1
(e-interaction) the c-actlOn
----
~-==-
User action
Interpretation
I User action
.-----1
IS actIOn [nterpret::n-l
act
I '----__----'I __ ~ct. _.,
Figure 6. JSP diagram showing the action structure of interactive usage situations
Ae-messages can be carried by, and visualized as, documents. We use

'document' as a generic concept referring to 'traditional' paper documents as
well as screen documents (e.g., forms, dialogues and windows) and, for
example, EDI documents. An interactive screen document is used to
interactively manipulate the IS and serves as an action medium in the
communication through the IS. In the bookshop example above, every Web
15 In reality this means of payment is probably rare. Most Web sites require credit details up
front, so that in fact the bank promises to pay on behalf of the customer at the time of
purchase, not the time of delivery.
page except for the last is an interactive screen document. The last page was
a static screen document used only to mediate the order acknowledgment to
the customer l6 •
The actability way of understanding IS usage is influenced by, and
should therefore be related to, the 'seven stages of action' model of Norman
(1988) as described in Section 3.1. Since e-actions and e-interactions are
both actions, but at different levels of abstraction, they can both be related to
the model, as shown in Table 4.
As Table 4 indicates, Norman's model has been extended with an
additional stage 4.5 to show the IS action performed within an interaction
loop. It is also during this stage that the possible consequences of the action
anse.
Table 4. Nonnan's (1988) 'seven stages of action' model compared with execution of e-
actions and fonnulation of ae-messages by e-interactions. 17
Sta!!" Norman <-actions e-interactions

il ? Forming the goal Deciding desired !
U A Forming the goal
-::----i r I perlocutionaryeffect(s) s c
+-;e Forming the intention Choosing action mode e t Choosing User action
3 ,p. SpecifYing the action Formulating the ae-message r i SpecifYing the User action.
4 ,IE Executing the action Executing the e-action 0 Performing the User action
--;-;:-i x n
4.5 Ie. Consequences arise IS action performed
5 !E Perceiving the system Perceiving the business state I a Perceiving the system's response
~v state n c
6 ia Interpreting the Interpreting the business state t I
t Performing the interpretation act
system state e
-ri
7 lu
a.
Evaluating the
outcome
Evaluating the performed
action and the IS
r
p.
Evaluating what the system has
achieved
In the above discussion, we have focused on interactive usage situations.

If a consequential usage situation includes interactive usage of an IS, many
of the same characteristics apply, i.e., interactions with the IS follow the
structure of the elementary interaction loop. What differs is that the
formulation is done implicitly with the support of the system instead of
explicitly in an interactive screen document. Furthermore, the execution and
evaluation phases are performed without using the system, i.e., manually. It
is important to understand that not every consequential usage situation
includes interactive usage of an IS. Instead, paper documents are often used
as carriers of messages in these situations.
16 That is, assuming that no action potential was made available by the page. It might very
well have been, for example, a button labelled 'print page'. In that case, it would have
been an interactive screen document.
17 Adopted and modified from Agerfalk (1999).
4.3 Actability features of information systems
From the above we can conclude that in order for an IS to be regarded as

actable during an interactive usage situation it should be able to help (permit,
promote and facilitate) an interacting performer to:
1. choose what communicative action to perform (at action level);
2. formulate the propositional content of the action and attach it to an
understandable and appropriate action mode (at interaction level); and
3. execute the communicative action; and finally to
4. interpret and evaluate the action and its business effect(s).
Furthermore, the IS should be able to help (permit, promote and
facilitate) an interpreter to acquire sufficient background to perform the
intended action during a consequential usage situation.
Finally, the IS should be able to make the communicator feel confident
that the performer (human or automaton) really performs the intended action
and that the resulting message reaches the intended interpreter(s).
Any usage situation must be understood within the context of doing
business with some other party. In a business setting, the interpreter
(recipient) of a 'business message' (ae-message) must trust what is said (and
thus communicatively done) by a speaker (communicator). To understand
the meaning of 'trust what is said' we tum to the 'universal validity claims'
proposed by Habermas (1984). He argues that such claims are raised by any
communicator and presupposed by the communicator to be accepted by the
interpreter. Successful communication implies that the listener must both
comprehend the action and accept it as valid. That is, the communicator and
the interpreter must agree on the communicative act in order for it to be
successful. To reach such mutual understanding a communicator raises four
universal validity claims (ibid.): (1) a claim of comprehensibility, (2) a claim
of truth, (3) a claim of rightness, and (4) a claim of sincerity.
The first claim, (1) the claim of comprehensibility, means that a
communicator must be able to formulate a grammatically correct,
comprehensible sentence. In the context of the bookshop example above this
means, for example, that the 'book of the month' on-line offer must be
expressed in a way understandable by potential customers. This can be
achieved in several different ways, for example by conforming to accepted
standards, protocols and user interface design guidelines. This relates to the
'permit, promote and facilitate' in the definition of actability (see above). It
is, for example, possible to talk about action visibility of an IS when its
action potential is 'self-explanatory', i.e., easy to understand and use (cf.
Norman (1988) on the concepts of 'affordance' and 'visibility'). The second
claim, (2) the claim of truth, means that an action must refer to the true state
of affairs. In the on-line offer example, this would, for example, imply that
there actually are books to be delivered as expressed in the offer and that
these books are correctly described. The third claim, (3) the claim of
rightness, means that a communicator must be able to establish an
interpersonal relationship in the right way according to accepted social
norms. For example, the bookshop must be able to act according to
applicable legal and cultural norms of trade. The fourth claim, (4) the claim
of sincerity, means that a communicator must have sincere intentions, for
example, that the books hop actually intends to sell its books to a visitor.
From this we can conclude that if an IS is to be regarded as actable its
users must be able to both comprehend it syntactically and semantically and
accept its underlying intentions as true, sincere and correct.
5. IMPLICATIONS FOR INFORMATION SYSTEMS

DESIGN
Adhering to the concept of actability has important implications for the

way information systems are designed. It implies a shift from the traditional
focus on data structures (i.e., syntactic and semantic aspects) to take the
systems, actors and their actions as the point of departure for design. It is by
understanding the structure of an organization's actions and hence its
communication patterns that it becomes possible to design actable
information systems. This also means a shift from focusing on the functions
of organizations and information systems. The action structures of an
organization form its business processes. Thus, it is not adequate to talk
about business functions that require an information system to have various
functionalities for it to be effective and efficient. Instead, information
systems design should focus on what the actors do (communicative and
material actions) to create value for the organization's clients. Designing an
IS then implies designing its action potential so that it matches the
requirements of its users. Designing for actability is a holistic approach
where the value-creating activities of an organization as a whole are
considered-there is no point in designing information systems isolated from
their action context.
From our work with methods to support actability-oriented requirements
engineering (e.g., Agerfalk, 1999) we have extracted two principles that we
believe are crucial for the design of actable information systems: (1) focus
on actions and (2) integrate early prototyping with visual modelling.
To (1) focus on actions means to use the concepts of e-action and ae-
message as a foundation for understanding the role and tasks of developed
information systems and for partitioning the development work into
manageable pieces. When designing the action structures of business
processes, the perfonnance of actions is allocated to human actors and

infonnation systems. From this model, it is straightforward to partition the
usage of infonnation systems into usage situations: interactive, automatic
and consequential. Each usage situation can then be analysed in tenns of
what e-actions are perfonned and what IS support is required to perfonn
them in tenns of interactions, interaction sequence restrictions and concepts
used to express them. To focus on actions means to always keep the
definition of actability in mind and think of how to design, so as to pennit,
promote and facilitate action by and through the system. One benefit of
focusing on actions is that requirements become traceable throughout the
different models created during systems development: from business models,
through interaction models, and down to detailed requirements on system
operations (Agerfalk, 1999). To focus on actions is also important in
understanding responsibilities within an organization. Who says what to
whom, when, and why. Identifying the communicators and stressing that
they are users even though they may never touch the system physically,
helps to shift the question of 'system ownership' into' action responsibility',
which is much more tangible for the persons involved.
(2) Early prototyping means creating examples of the user interface early
in the development process. The user interface is the visual part of the IS
with which interacting users communicate. Exploring different alternatives
together with the users is essential to design the visual layout of the system.
Additionally, it is an invaluable tool to understand the professional language
of the users and hence an important input to conceptual modelling. Our
research has also shown (Agerfalk, 1999) that creating prototypes helps, not
only in getting the system right, but also in gaining an understanding of how
the business will actually be perfonned. Hence, user interface proto typing is
an important input to business modelling as well. By visual modelling, we
refer to analytical descriptions of a system in some defined notation (e.g.,
class diagrams and data flow diagrams). Visual models are important for
promoting communication about a system at different levels of abstraction.
To understand a system under construction, developers should both
exemplify (with prototypes) and analytically abstract (with visual models).
These two approaches can be integrated to utilize the strengths of both (cf.
Mathiassen et aI., 1995). Prototypes are important, for example, to discuss
interactive screen documents with users in order to gain knowledge
important to creating the visual models. On the other hand, visual models
can serve as important 'road maps' to help users focus on different parts of
the system when the prototypes are being created (Agerfalk, 1999).
6. CONCLUSIONS
There is an urgent need to create a deeper understanding of information

systems in organizational settings. Information systems are technically
implemented, but they cannot be understood as mere technical objects. Their
special character is that they are formalized sign systems and as such are
used for human communication. A semiotic perspective, emphasizing the
sign and communication character of information systems, is necessary
when establishing an organizational theory of information systems. In this
chapter we have investigated different aspects of information systems from a
semiotic standpoint with emphasis on the top stages of the semiotic ladder
(Stamper, 1994)-the social and pragmatic aspects. A key concept for such
an understanding of social and pragmatic aspects of information systems has
been presented: information systems actability. Standing on the top of the
ladder, we have also discussed some matters concerning meaning, structure
and media.
We have identified two different types of actions related to signs:
Action within the sign transfer
- Action based on the transferred sign
These two types of actions can be related to our conceptualization of
information systems actability and the three types of IS usage situations.
Action within sign transfer is what is performed during interactive and
automatic usage situations. Action based on the transferred sign is what is
performed during consequential usage situations and sometimes during
interactive usage situations.
We have also analysed the role of messages (sign constellations) within
information systems. In our analysis we have used speech act theory as our
main source of inspiration. This has led us to the concept of ae-messages
consisting of an action mode (type of illocution) and a propositional content.
Ae-messages are results of communicative actions that we call e-actions.
Input and output messages of information systems are realized through
documents. Interactive screen documents playa special role: not only are
they a result of action, they are also used in interactive usage situations for
several purposes. Such a document can aid the user to formulate the ae-
message (the propositional content) and attach it to an appropriate action
mode. In addition, it is also an action medium for the execution of the e-
action.
We claim information systems actability to be an important
conceptualization for understanding the use of information systems in
organizations. It makes it possible-within one theoretical framework-to
relate different aspects (such as technical, cognitive, behavioural, linguistic
and organizational) to each other. This is possible because of our basic
pragmatic perspective. Starting from human action, we have given a

coherent description of social action, organizational action and artefacts
usage as a basis for formulating the concept of information systems
actability.
REFERENCES
Agerfalk, P. J. 1999, Pragmatization ofInformation Systems - A Theoretical and
Methodological Outline, Licentiate thesis, IDA, Linkoping University, Sweden.
Agerfalk, P. 1., Goldkuhl, G. and Cronholm, S. 1999, Information Systems Actability
Engineering: Integrating Analysis of Business Processes and Usability Requirements, in
Proceedings, 4th International Workshop on the Language Action Perspective on
Communication Modelling LAP'99, pp.73-86, Copenhagen, Denmark, September 12-13,
1999.
Ahrne, G. 1994, Social organizations. Interaction inside, outside and between organizations,
Sage, London.
Austin, J. L. 1962, How to do things with words, Oxford University Press, Cambridge.
Berger, P. L. and Luckmann, T. 1967, The social construction of reality, Doubleday & Co,
Garden City.
Cronholm, S., Agerfalk, P. J. and Goldkuhl, G. 1999, From Usability to Actability, in
Proceedings, 8th IntI. Conference on Human-Computer Interaction HCI International '99,
Munich, August 22-27, 1999.
Dietz, 1. L. G. 1994, Business modelling for business redesign, in Proc. 27th Annual Hawaii
IntI. Conference on Systems Science, IEEE.
Giddens, A. 1984, The constitution of society. Outline of the theory of structuration, Polity
Press, Cambridge.
Goldkuhl, G. 1995, Information as action and communication, in Dahlbom, B. Ed., 1995, The
infological equation. Essays in honor of Borje Langefors, Gothenburg Studies in
Information Systems 6, Goteborg University.
Goldkuhl, G. 1996, Generic business frameworks and action modelling, in Proceedings
Communication modelling - Language/Action Perspective '96, Springer Verlag.
Goldkuhl, G. 1998a, Handlingars vad, vern och var. Ett praktikteoretiskt bidrag, in Swedish"
Working paper, CMTO, Linkoping University.
Goldkuhl, G. 1998b, Towards an action oriented understanding of organizational change,
Working paper, CMTO, Linkoping University.
Goldkuhl, G. and Agerfalk, P. J. 1998, Action within Information Systems: Outline of a
Requirements Engineering Method, in Proceedings 4th International Workshop on
Requirements Engineering: Foundation for Software Quality REFSQ'98,. Pisa, Italy, June
8-9,1998.
Goldkuhl, G. and Lyytinen, K. 1982, A language action view of information systems, in
Proceedings 3rd International Conference on Information Systems, Ann Arbor.
Goldkuhl, G. and Nilsson, E. 2000, Organisational Ability - constituents and congruencies,
accepted for OR42, Swansea, Wales
Goldkuhl, G. and Rostlinger, A. 1999, Expanding the scope: From language action to generic
practice, in Proceedings 4th International Workshop on the Language Action Perspective
on Communication Modelling LAP'99,. Copenhagen, Denmark, September 12-13,1999.
Habennas, J. 1984, The theory of communicative action 1. Reason and the rationalization of
society, Polity Press, Cambridge.
Joas, H. 1993, Pragmatism and social theory, University of Chicago Press, Chicago.
Langefors, B. 1966, Theoretical analysis of infonnation systems, Studentlitteratur, Lund.
Langefors, B. 1995, Essays on Infology, Dahlbom B. ed., Studentlitteratur, Lund, Sweden.
Latour, B. 1992, Technology is society made durable, in Law ed., 1992, A sociology of
monsters: essays on power, technology and domination, Routledge& Kegan Paul, London.
Liu, K., Stamper, R., Clarke, R. and Andersen, P. B. eds., 200 I, Infonnation, Organisation
and Technology: Studies in Organisational Semiotics, Kluwer Academic Press, Boston.
Mathiassen, L., Seewaldt, T. and Stage, J. 1995, Prototyping and SpecifYing: Principles and
Practices of a Mixed Approach, Scandinavian Journal ofInfonnation Systems, Vol. 7, No.
1.
Nonnan, D. A. 1988, The psychology of everyday things, Basic Books, New York.
Preece, J., Rogers, Y., Sharp, H., Benyon, D., Holland, S. and Carey, T. 1994, Human-
computer-interaction, Addison-Wesley.
Scott, W. R. 1995, Institutions and organizations, Sage, Thousand Oaks.
Searle, J. R. 1969, Speech acts. An essay in the philosophy oflanguage, Cambridge
University Press, London.
Stamper, R. K. 1994, Signs, infonnation, nonns and systems, in Holmqvist, B., Andersen, P.
B., Klein, H. and Posner, R. eds., 1994, Signs at work, De Gruyter, Berlin.
Stamper, R. 2001, Infonnatics without the Computer, in Liu, K., Stamper, R., Clarke, R. and
Andersen, P. B. eds., 2001, Infonnation, Organisation and Technology: Studies in
Organisational Semiotics, Kluwer Academic Press, Boston.
Thayer, H. S. 1981, Meaning and action. A critical history of pragmatism, Hackett Publishing,
Indianapolis.
Weber, M. 1978, Economy and society, University of California Press, Berkeley.
Weizenbaum, J. 1976, Computer power and human reason, Freeman, San Francisco.
Winograd, T. and Flores, F 1986, Understanding computers and cognition: A new foundation
for design, Ablex, Norwood.
Chapter 5
Communication Quality in the Context of Information

Systems and Business Processes
Owen Eriksson
1. INTRODUCTION
There is a growing interest in communication and social interaction in the

business area. New marketing theories (e.g. Hakansson and Snehota, 1995;
Gummesson, 1995; Gronroos, 1996; Zeithaml, 1990) stress the importance
of business relationships, interaction and communication for the business
process. The rapid development of information technology, especially in the
area of data communication has also created a growing interest for using
information technology (IT) and information systems (IS) in the horizontal
business communication between suppliers and customers. Sales support
systems, agent business systems and electronic commerce on the Internet are
becoming more and more important for business communication between
companies, organisations and consumers. New perspectives and theories for
analysing business processes and business communication have also
emerged during the last ten years in the information sciences research area,
where the importance of communication for the business process is stressed
(e.g. Auramiiki et. al. 1992; Chong and Liu, 2001, Dietz, 1994; Denning and
Medina-Mora, 1995; Filipe et.al., 2001; Goldkuhl, 1998; Stamper, 1994).
The purpose of the paper is to discuss the importance of communication
and the concept of communication quality in the context of business
processes and information systems (Eriksson, 1999). The basic idea behind
116 Owen Eriksson
the concept of communication quality is that high quality communication is

equal to communication with qualities that contribute to mutual
understanding. This idea is grounded in speech act theory (Austin, 1962;
Searle, 1969, 1979), formal pragmatics (Habermas, 1979), communicative
action theory (Habermas, 1984) and organisational semiotics (Stamper,
1994).
2. SPEECH ACT THEORY, FORMAL PRAGMATICS

AND COMMUNICATIVE ACTION THEORY
Austin (1962), the language philosopher who developed speech act

theory, claimed that we cannot understand what is meant by a meaningful
and successful use of language if we only think that it is used to present facts
about the world that can be either true or false. Austin claimed that we use
language to perform speech acts (illocutionary acts) e.g. to promise, ask,
order, warn, request, etc. that could be evaluated either as happy or unhappy.
Searle (1969, 1979) was Austin's student and one of Searle's (1969)
major contributions to speech act theory is his analysis of general
presuppositions for successful speech acts. The result of Searle's analysis is
formulated in four general rules: propositional content rule, preparaty rule,
sincerity rule and essential rule, which are rules that the speaker must
comply to if she/he wants to succeed with his/her speech act.
Habermas (1979) is a language philosopher and sociologist who has
developed speech act theory into a theory of formal pragmatics. Habermas
(1979, 1984) claims that the main purpose of formal pragmatics is to analyse
universal presuppositions for successful communication. The universal rules
imply that a speaker who performs a speech act, and who is oriented towards
mutual understanding, must raise universal validity claims:
comprehensibility, sincerity, truth and normative rightness and presuppose
that they will be accepted by the listener. This means that the quality of the
speech act is dependent on communicative rationality and validity.
Successful communication implies that the listener must both comprehend
and accept the speech act, that is the speaker and listener must agree about
the speech act. If the speech act is accepted in this wayan actor relationship
based on mutual understanding is created, and that implies obligations that
regulate the social interaction. Habermas' interest in speech act theory is due
to his aim to analyse the importance of human communication for social
interaction. In the theory of communicative action, social interaction is
defined as action contexts where two or more social actors co-ordinate their
material and speech acts to carry out their plans of action (Habermas, 1984).
The main focus of Habermas is to analyse the mechanisms that allow people
Communication Quality in the Context of Information Systems... 117
to co-ordinate their actions. The concept of social interaction can be

illustrated if we imagine a teacher that performs the following speech act in a
classroom:
I request that you open the window (to the student sitting next to the
window).
The teacher's objective is to get some fresh air into the classroom, the
plan of action is that the student should perform the material act to open the
window; the teacher wants to carry out the plan by co-ordinating his actions
with the student, and the means of this co-ordination is the speech act.
Notice that the social interaction includes both the speech act (the request),
and the material act of opening the window.
3. COMMUNICATION QUALITY
A very important distinction that is made in speech act theory and formal
pragmatics is the difference between the illocutionary aim and the
perlocutionary effect of the speech act (Searle 1969, pp. 46-47; Habermas
1979), and this distinction is very important for the concept of
communication quality. The distinction can be illustrated by the teacher's
request to open the window:
1. the illocutionary effect implies that the speaker (the teacher) can get the
listener (the student) to understand what he means;
2. the perlocutionary effect implies that the listener (the student) will open
the window.
This means that the speech act is considered to be successful when the
listener has understood the request. The reason why high quality
communication should be evaluated primarily from the iIIocutionary
aim/effect is that language is primarily a means for creating understanding
between human actors. This can be illustrated by the teacher's request. The
teacher does not open the window with his speech act; he actually makes a
request to open the window, which is an important difference. This implies
that the teacher, with his speech act, must create an understanding of what he
wants the student to do, and to motivate the student to cooperate. The teacher
must be able to create this mutual understanding because the student is an
independent actor that must understand, and can accept or reject the request.
It is in this context that Searle's general rules and Habermas validity claims
are important. The idea with speech act theory and formal pragmatics is that
successful communication is based on mutual understanding, which is
dependent on social conventions, the validity of the speech act and actor
relationships. This implies that communication quality can be recognised as
communication with qualities that contribute to actor relationships based on
118 Owen Eriksson
mutual understanding. But this does not mean that the perlocutionary effects
and efficacy of the communication is unimportant. Speech acts are used in
social interaction, which implies that the actors use speech acts to coordinate
their goal-directed actions, and this means that the actors also must be able
to perform communication and material acts in an efficient way. From a
social interaction point-of-view it is essential that the student also succeeds
in opening the window, and perlocutionary effects are also important for the
quality of the communication. Still it is important to emphasise the generic
importance of illocutionary aim/effects and mutual understanding for the
quality of the communication. Ifwe were only to evaluate the speech act for
perlocutionary effects it would not be possible to distinguish a deceptive
speech act from a non-deceptive one. For example, the teacher can make the
student open the widow both with an honest and dishonest request, and if the
opening of the window was the only thing that mattered it would not matter
whether the teacher deceived the student into making the material act or not.
The important distinction between illocutionary and perlocutionary
effects is also discussed by Stamper (1994). He says, "The difference
between iIIocution and perlocution is obviously important - you experience
this when you pass the man with the placard declaring "The end of the world
is nigh!" The perlocutionary response is seldom what we intended". Stamper
has presented a semiotic framework called the semiotic ladder, which
describes different levels of structures, meanings and usage of signs. In this
framework the concept of pragmatic meaning is important, which is related
to the iIIocutionary intended effect of the speech act. Stamper (1994) also
discusses the importance of shared norms for communication and pragmatic
meaning. He claims that a sign is meaningful if it relates to, or changes, a
norm structure. This is also an important idea in speech act theory, that is the
pragmatic meaning of speech acts is dependent on social norms, rules and
conventions. The concept of communication quality can be related to the two
highest levels of the semiotic ladder which deals with the pragmatics of
signs and the social world, and foremost to the concepts pragmatic meaning
and social norms (Stamper, 1994). According to Habermas (1984) is the
pragmatic meaning of the speech acts and social norms is dependent on
shared values, validity and mutual understanding between social actors.
4. BUSINESS PROCESSES: A TYPE OF SOCIAL

INTERACTION
The business process can be characterised as a type of social interaction

where material and communication acts are performed. The reason why I use
the concept communication act instead of speech act, is that communication
Communication Quality in the Context of Information Systems ... 119
can be performed with other media than spoken language, for example,
paper and electronic documents produced by an IS. Social interaction can be
described in two steps:
Step A: the actors make a common interpretation of the action context
and the results that are to be achieved;
Step B: the actors perform actions to realise action plans, goals and
results.
The social interaction that takes place in business processes is illustrated
in Figure 1 below.
Action context
Step 1: a common Interpretadon or the acdon context
Interpreter Sender
Sender Interoreter
Inte reter Sender

Purchase order
Sender and interpreter L..,-----~------;TSeDder and interpreter
Purch.. e contract
Salelman ctor re a ODS P w e i aled on an Customer

agreement that Impllel loclal commitments
and expectadonl and a common knowledge
wblch overnl the actlonlln Ite 2
Step 2 I'uUllment or agreementl and commItments
DeUvery
Payment "") )
.--------------------------------
SlIlesman
BUllne.. relatlonlblp whIch II based on how the acton create
agreement. and rulfill mutual commItments
I
Customer
Figure 1. Business processes a type of social interaction
Figure 1 shows that the actors in step A make a common interpretation of

the action context, which results in a business contract (business agreement).
Communication acts are necessary to establish this agreement and to create
an actor relationship, which governs the subsequent actions in the business
process (in step B). The figure shows four basic communication acts that are
used create the business agreement in step A.
1. The customer puts forward a request or a wish for a product that he/she
wants. From a communication action perspective the customer is the
sender and the salesman is the interpreter.
120 Owen Eriksson
2. The salesman makes a business offer where the salesman is the sender
and the customer is the interpreter.
3. The customer makes a purchase order where the customer is the sender
and the salesman is the interpreter.
4. The purchase order and its acceptance are confirmed in a purchase
contract where the customer and the salesman both are sender and
interpreter at the same time.
In step B actions are performed to fulfil the business agreement. The
salesman is responsible for delivering a product and the customer must
receive and pay for the product. These actions are governed by an actor
relationship that consists of:
- mutual commitments;
common knowledge about the action context, which includes knowledge
about the results that should be achieved;
- social expectations.
The description that I have made above is of course a simplification.
Social interaction is a teleological language game, which implies that the
interpretation of the action context in step A is not only dependent on
communication, but also on material acts and empirical observations. For
example: the customer can also feel and look at the car to evaluate and
decide whether she/he wants to buy it. The description of step B is also a
simplification. There are not only material acts and payments that are
performed in step B. For example the salesman must communicate and make
new agreements with other actors that will actually have the job of
delivering the car. This means that both steps A and B can be further divided
in a number of phases as shown in the BAT-model (Goldkuhl, 1998).
Goldkuhl (1996, 1998) has presented the BAT-model (Business Action
Theory model) that describes the business process as social interaction. The
BAT-model is a generic business framework where the business transaction
is described as an interchange process between supplier and customer, which
involves the creation and sustenance of business relationships. The
framework is divided into six phases:
1. Business requisite phase.
2. Exposure and contact search phase.
3. Contact establishment and proposal phase.
4. Contractual phase.
5. Fulfilment phase.
6. Completion phase.
Step A, which was described in figure 1 above, can be related to phase 3
and 4 in the BAT-model, and step B can be related to phase 5. The purpose
of the BAT-model is to describe the inherent business logic and generic
business acts in the business process. Generic business acts consist of both
material and communication acts. An important aim with the BAT-model is

to emphasise the importance of communication for the business process,
Goldkuhl (1997, p 5) claims that
"Making business involves with necessity communication otherwise
customer and supplier cannot agree on the business deal. But business
interaction cannot be reduced to only communication. It must include the
material acts of delivering goods and/or services and paying."
5. A SALES SUPPORT SYSTEM AND THE CAR

SELLIPURCHASE PROCESS
In the previous section I have claimed that the business process:

- can be described as a type of social interaction;
- can be divided into different phases according to the BAT-model.
Furthermore, it is important to emphasise that information systems are
used to perform important communication acts in the context of business
processes. This can be illustrated by a sales support system that is used by
the Volvo/Renault car dealers in Sweden. The sales support system is used
by the car dealers in interaction with the customers at the car dealer's office.
I have empirically evaluated the system during a period of 1.5 years, and the
results of this evaluation process have been presented in (Eriksson, 1996;
1998). How the system is used in the car sell/purchase process can be
described in relation to the contact and establishment, contractual and
fulfilment phases of the BAT-model, according to the description below.
5.1 A sales support system and how it can be used in the

activities of the car sell/purchase process
5.1.1 Contact establishment and proposal phase
The contact establishment and proposal phase consists of the following

activities: demonstration, negotiation and purchase decision.
Demonstration
In the demonstration activity the system can be used by the car dealer
together with the customer to create an image of a new car on the computer
screen by combining different car models with different options. The
customer can also choose extra equipment with the help of the system.
Negotiation
In this activity the system can be used to present detailed prices, to find
and record information about the customer's trade-in car. If the purchase is
122 Owen Eriksson
to be paid by instalments or if the car is to be leased the system can be used

to calculate the instalments and leasing fee. The car dealer can communicate
a detailed business offer to the customer with the help of the system.
Purchase decision
When the customer has received an offer he/she can make a purchase
decision, which often is made at home or at the customer's work-place. The
purchase decision is often made together with other actors for example
together with the family.
5.1.2 Contractual phase
The contractual phase consists of the activities of purchase order and

contracting.
Purchase order
If the customer wants to buy the car he/she gives an purchase order. The
customer will either come to the car dealer's office to do this, or she/he will
call the car dealer by telephone.
Contracting
When the customer has made his/her purchase order the car dealer can
use the system to present a purchase contract that must be signed both by the
customer and the car dealer.
5.1.3 Fulfilment phase
Internal ordering
The car dealer InItiates the fulfilment phase by communicating an
internal order to the delivery office. The clerk at the delivery office uses the
internal order to order the car from the general agent and the factory.
Delivery
In the fulfilment phase the car dealer is responsible for delivering the car
that is specified in the purchase contract.
Payment
In the fulfilment phase the customer is responsible for paying for the car.
This payment can be made in different ways. The customer can pay for the
car at once or he/she can pay by instalments or by leasing fees.
5.2 The sales support system can be considered as a

vehicle and agent for business communication
The example above shows that the sales support system is used by the car
dealer to perform important business actions, illustrated in Figure 2 below.
A<tor relationshIp
Commonlcatlon a<t Interpretation
Relation between
Re1adonb_ the actor an" IS
the a<:tor an" IS Information system
Sen.... Redplent PerConner Interpreter
Figure 2. A communication action perspective of IS
The figure shows that the car dealer performs a business act, with the
help of the sales support system, which results in a message that changes and
affects the relationship between the car dealer and customer. The business
offer is important for how the actors interpret the action context and it
creates social commitments and expectations that change the status of the
actor relationship. The sales support systems can be regarded as a vehicle for
communication because the system is used to perform communication acts
that are expressed with a message (Agerfalk, et. aI., 1999). But the system
can also be considered as a social agent (ibid.). The IS can either play the
role of the performer in relation to the interpreter, or of the recipient in
relation to the sender. The reason why IS also can be regarded as a social
agent, and not only as a vehicle for communication, is that:
- In an autonomous way IS can receive messages from the sender, this
implies that IS can take the role of a receiver in relation to the sender;
- In an autonomous way IS can send messages to the interpreter, this
implies that IS can take the role of a performer in relation to the
interpreter.
6. COMMUNICATION QUALITY IS A GENERIC

QUALITY CONCEPT IN THE CONTEXT OF
BUSINESS PROCESS AND INFORMATION
SYSTEMS
The example described in the previous section illustrates why

communication quality is a generic quality in the context of information
systems and business processes. The reasons for this are that:
124 Owen Eriksson
information systems can be described both as a vehicle and social agent

for business communication;
business communication is used to perform generic business acts:
request, offer, order or contract;
the communication acts create business agreements and actor
relationships that imply commitments and obligations that must be
fulfilled in the course of the business process.
This means that the ability to create business agreements and to fulfil
these agreements is generic for business processes, which implies that it is
very important to be able to create agreements of a high quality and to fulfil
these agreements and commitments in a reliable way. This also means that
communication that is performed with the help of IS changes the business
communication and the quality of the communication. For example, it is
important that the business offer, shown in Figure 3, is of high quality.
Date: 18-12-97 Time: 10:30

OfTer
Sales man: BoEk
Customer: Eva Larsson
Purchase object:
Model: Volvo 850 GLT 2,5. Front-wheel drive, 4 doors.
Engine: Petrol. Catalytic. 5-cyIinder. Transverse. 170 h.p.l125 kW B5254F
Gear Box: Manual 5-geared. M56.
Color: Polar white nr: 189
Extra Equipment: S-Package: Automatic climate control + Cruise control.
Total: 241,000 SEK.
Salesman's signature. Period of validity 31-12-97
Figure 3. A Business offer
My basic definition of the concept of communication quality was

presented in section 3 above, and it says that communication quality can be
recognised as communication with qualities that contribute to actor
relationships based on mutual understanding. This means that it is important
that the business offer above contributes to a car dealer-customer
relationship built on mutual understanding. The business offer above can be
described as an action-elementary message (Agerfalk. et. aI., 1999) that
consists of a propositional content and an action aspect, which indicates the
communicative function of the message. It is important that both the action
aspect and the information content of the message are of high quality. In the
message above the information content describes the attributes of the car.
The action aspect of the message, which is expressed with the iIlocutionary
verb offer in the example above, shows how the information content should
be used. The action aspect also indicates the status of the business
relationship. The message must also be comprehensible and trustworthy
which means that the language and the symbols that are used must be
understood, and that the customer can depend on it, which implies that the
customer should be able to control and criticise the business offer, and that
the car dealer is able to defend, explain and to perhaps modify the offer if it
is criticised. It is important that the business offer contributes to a valid
argumentation. If the customer later on makes a purchase order and if the car
dealer accepts it, a purchase contract is signed. The purchase contract
confirms the business agreement and implies a stronger commitment when
compared to the business offer. The car dealer has made a commitment to
deliver the car, and the customer has made a commitment to receive the car
and to fulfil the payment. This implies that communication quality built on
mutual understanding is very important in the context of business processes,
and that IS should contribute to this quality. It is important to emphasise that
the mutual understanding of the business agreement is not only restricted to
a mutual understanding of the results that should be achieved. Both in speech
act theory and organisational semiotics it is argued that the pragmatic
understanding of signs and messages are related to social norms, values and
behaviour. For example, when the actors have signed the purchase contract it
is important that both the actors understand that they have made a fair
business agreement, a judgment that is to a high extent based both on formal
and informal social norms and values. It is also important that the actors, in
their subsequent actions, behave according to the specific norms that are
referred to in the purchase contract. The actors must also act according to
more general and informal norms which you would expect from reliable and
social actors. Reliable business partners should for example fulfil their
promises and commitments. Agreements and commitments are in general
related to social norms and have a social binding force that function as
guidelines for the actions that are performed; actions are, to a high extent,
norm regulated (Habermas, 1984). Stamper (1994) says "A norm is more
like a field of force that makes the members of the community tend to
behave or think a certain way." Trustworthiness and reliability are also
things that have been discussed in the area of service management, where the
concept of service quality has been studied for a long time. Gronroos' (1990)
discusses the importance of the supplier making correct promises, making
the promises in a correct way and fulfilling the promises and commitments
made. Zeithaml et. al. (1990, pp. 116-118) claim that two important factors
that can affect service quality in a negative way are inadequate horizontal
126 Owen Eriksson
communication and a propensity to over promise. Both Granroos and

Zeithaml et. al. emphasise that it is important that the supplier is able to give
trustworthy promises and that the supplier has to fulfil these promises in a
reliable way. This also implies that communication and communication
quality must be evaluated in a social interaction context, which may include
material acts. The reason why also material acts have to be considered when
we analyse communication quality is that:
- communication acts function as coordinating mechanisms for material
acts (Habermas, 1984);
- material acts also have a communicative and social meaning.
This implies that communication quality in the context can be defined as:
Communication quality in the context of information systems and
business processes are qualities of communication and material acts that
contribute to business relationships that are built on mutual understanding
and trust. Communication and service quality are two concepts that have
much in common but the problem with the concept of service quality, from a
communication quality point-of-view, is the focus on the supplier. It is the
promises of the supplier and the supplier's obligation to fulfil his promises
and commitments that are emphasised. This is different to the
communication quality point-of-view where the mutuality and social aspects
of promises and commitments are stressed. This implies that communication
affects actor relationships and the social world and that communication
quality is a social quality. The point with the concept of communication
quality is that it is focused on the mutuality of interaction and relationships.
The reasons for this are that communication is a social and mutual concept
and that both the customer and supplier make promises and commitments
that must be fulfilled in the course of the business process.
I claim that communication quality is a generic quality in the context of
business processes and that IS that are used for business communication
affect and should contribute to this quality. Communication quality is a
theoretical concept that has been used to evaluate business processes and
information systems used in car deals, and in the tourism industry (Eriksson,
1996, 1998, 2000). These empirical case studies show that communication
quality is an important quality in the context of information systems and
business processes.
7. CONCLUSIONS
I have claimed that the business processes can be regarded as a type of

social interaction and that information systems are used to perform important
business actions in the context of business processes. I have also introduced
the concept of communication quality in the context of information systems

and business processes. Quality in the context of business processes is often
described either in terms of more or less objective qualities of physical
products (Bergman and Klefsjo, 1994), or in terms of the customer's
subjective experiences (Gronroos, 1996; Zeithaml et. aI., 1990). This is
different from the concept of communication quality where the
communicative, social and mutual aspects of quality are focused. The
concept is focused on the importance of actor relationships and social norms
i.e. the social world for the quality of business processes. But
communication quality is also an important presupposition for efficacy and
material quality in business processes; for example, high communication
quality is a presupposition if both the customer and supplier are to be able to
achieve the results that they are striving for. I also claim that IS affects
communication quality and that communication quality is a type of quality
that becomes more and more important for different actors e.g. salesmen,
suppliers and customers who are involved in the actual process of doing
business. One important reason for this is that information technology is
becoming more and more used in the process of doing business, e.g. the
concept is very important in the context of electronic commerce on the
Internet.
REFERENCES
Auramaki E., Hirschheim R., Lyytinen K. 1992, Modelling Offices Through Discourse
Analyses: A comparison and Evaluation of SAMPO with OSSAD and ICN, The Computer
Journal, Vol. 35, No.5 pp. 492-500
Austin J.L. 1962, How to do things with words, Oxford University Press, UK, Eds. J.O.
Urmson, M. Sbisa
Bergman B. Klefsjo B. 1994, Quality: from customer needs to customer satisfaction,
Studentlitteratur, Lund
Chong S., Liu K. 2001, A Semiotic Approach for Distinguishing Responsibilities in Agent-
Based Systems, In: Liu K., Andersen P. B., Clarke R. 1., Stamper R. Eds., 2001
"Information, Organisation and Technology: Studies in Organisational Semiotics",
Kluwer Academic Press, Boston.
Denning P.J. Medina-Mora R. 1995, Completing the loops, Interfaces, Vol. 25, No.3, pp. 42-
57
Eriksson O. 1996. A Communicative Action Analyses ofIn formation Systemes, Accepted to
the International workshop on Language/Action Perspective Tilburg, The Netherlands,
July 1-2, 1996
Eriksson O. 1998, Communication Quality: A sales Support System in The Eyes of the
Customer, T & S Working Paper 1998:5, Dalarna University, Borlange, Sweden, Accepted
to. Beyond Convergence ITS98 twelfth biennial conference, Stockholm, Sweden, June 21-
24, 1998
128 Owen Eriksson
"Infonnation, Organisation and Technology: Studies in Organisational Semiotics ",

Kluwer Academic Press, Boston.
Denning P.J. Medina-Mora R. 1995, Completing the loops, Interfaces, Vol. 25, No.3, pp. 42-
57
Eriksson O. 1996. A Communicative Action Analyses ofIn fonnation Systemes , Accepted to
the International workshop on Language/Action Perspective Tilburg, The Netherlands,
July 1-2, 1996
Eriksson O. 1998, Communication Quality: A sales Support System in The Eyes of the
Customer, T & S Working Paper 1998:5, Dalarna University, BorHinge, Sweden, Accepted
to. Beyond Convergence ITS98 twelfth biennial conference, Stockholm, Sweden, June 21-
24,1998
Eriksson O. 1999, Communication Quality: Towards an Intersubjective Understanding of
Quality, Accepted to: Quergo'99 International Conference on TQM and Human Factors,
Linkoping, Sweden, June 15-17, 1999
Eriksson O. 2000, Kommunikationskvalitet hos infonnationssystem och affarsprocesser, In
Swedish, Linkopings universitet
Filipe J., Liu K., Sharp B. 2001, Can Agents Model Organisation Roles?, In: Liu K.,
Andersen P. B., Clarke R. J., Stamper R. Eds., 2001, "Infonnation, Organisation and
Technology: Studies in Organisational Semiotics ", Kluwer Academic Press, Boston.
Goldkuhl G. 1996, Generic Business Frameworks and Action Modelling, Accepted to the
International workshop on Language/Action Perspective, Tilburg, the Netherlands, July 1-
2, 1996, Department of Computer and Infonnation Science, Linkoping University, Sweden
Goldkuhl G. 1997, The six phases of business processes - further development of Business
Action Theory, Centre for studies on human, technology and organization CMTO,
Linkoping University
Goldkuhl G. 1998, The six phases of business processes - Business Communication and the
Exchange of Value, Working Paper 1998-3, JOnkoping International Business School
Habennas J. 1979, What is Universal Pragmatics?, In: Communication and the Evolution of
Society, London: Heinemann.
Habennas l 1984, The Theory of Communicative Action, Volume One, Reason and the
Rationalization of Society, Beacon Press, Boston
Gronroos C. 1996, MarknadsfOring i tjanstefOretag, In Swedish, Liber-Hennods, Malmo
Gummesson E. 1995, Relationsmarknadsf6ring: Fnin 4P till 30R, In Swedish, Liber-
Hennods, Malmo
Hakansson H., Snehota I. Eds., 1995, Developing Relationships in Business Networks,
International Thomson Business Press, London
Searle J.R. 1969, Speech Acts - an Essay in the Philosophy of Language, Cambridge
University Press, London
Searle J.R. 1979, Expression and Meaning, Cambridge University Press, Cambridge
Stamper R. 1994, Signs, Infonnation, Nonns and Systems, In: Holmqvist B., Andersen P.B.,
Klein H., Posner R. Eds, Signs at Work, De Gruyter, Berlin, pp. 349-397
Zeithaml V. Parasuraman A. Berry. Leonard L. 1990, Delivering quality service: balancing
customer perceptions and expectations. New York Free Press.
Agerfalk P.l Goldkuhl G. Cronholm S. 1999, Infonnation Systems Actability Engineering-
Integrating Analysis of Business Processes and Usability Requirements, Accepted to: The
Fourth International workshop on the Language Action Perspective on Communication
Modelling LAP'99, Copenhagen, September 12-13, 1999
Chapter 6
Greimas' Semiotics and the Analysis of Organisational

Action
Daniel Robichaud
Over the past 20 years, an increasing number of organisational theorists

interested in understanding the constitution of organisation have shifted their
attention from the study of organisational structures to the analysis of the
interaction processes through which organisations are constituted and
maintained over time (Weick, 1979, 1995). This change was not a mere
change of conceptual strategy, but a change in the definition of the problem
itself. Rather than being taken for granted, the organisation and its structural
dimensions became the very phenomenon to be explained (see Weick, 1979,
1995; Putnam and Pacanowski, 1983; Poole and McPhee, 1983; Taylor,
1993; Taylor and Van Every 1999). In that vein, a wide variety of
perspectives were put forward to describe the organising processes by which
organisations emerge out of interaction and are reproduced in the course of
daily routines. These included systemic, interpretive or phenomenological
(Berger and Luckmann, 1966; Putnam and Pacanowski 1983) and
structurational perspectives (Ranson et at, 1980; Giddens, 1984). Among the
metaphors used to describe organising, that of a grammar suggested by
Weick has been particularly influential (Weick, 1979, 1995; Bantz, 1989;
Taylor and Van Every, 1999). Such a metaphor drove researchers' attention
from the content of organisational activities to the implicit rules and schemas
involved in organising.
The purpose of this chapter is to illustrate how semiotic theory can
contribute to the conceptualisation of organisational action. More
130 Daniel Robichaud
specifically, I would like to illustrate how the work of the French semiotician
Algirdas Julien Greimas in elaborating a narrative grammar shed a light on
organisational activities and their structuring aspects. The basic assumption
underlying this chapter is that Greimas' textual semiotics can be usefully
apply to the analysis of organisational action, and not only to its
representation in texts, the traditional object of study of semiotics. Drawing
on my extensive field research in a Canadian municipal administration, the
essay grounds this theoretical argument in an analysis of the organising of a
public discussion process considered as a narrative program (Greimas,
1983).
This chapter is divided in three parts. First, I briefly describe Greimas'
narrative semiotics and its core notion of narrative program. The second
section is devoted to a presentation of the organisational episode I use as an
empirical case in this chapter. In the third and fourth parts, I propose and
discuss a narrative analysis of the episode contending that semiotic theory
can provide us with a new perspective on organisational action. Specifically,
I argue that such a semiotic perspective ties together the idea of human
agency and that of structural constraints in a more satisfactory way than
other contemporary perspectives on the constitution of organisation.
1. GREIMAS' NARRATIVE GRAMMAR
Certainly the most prominent figure of what has been called the "Paris
Semiotic School" (Perron and Collins, 1989), A. J. Greimas is well known
for his contribution to semiotic theory. He argued that meaning relies at least
in part on basic structures that are to be found underneath the variety of their
manifestations in texts or cultural artifacts. Among the structural forms that
struck Greimas as central to the processes of meaning were the structures of
narratives. Following Propp's study of Russian folktales (Propp, 1968) and
Levi-Strauss' investigations of the structures of myths (Levi-Strauss, 1958),
Greimas developed a narrative grammar based on the hypothesis that
narrative structures were largely responsible for the meaning of any text. He
argued that narrativity was at the core of the organisation of meaning, since
it defines a set of structural relationships between actors and objects
involved in a course of action. He generalized the notion of narrativity,
contending that it cannot be reduced to a property of a particular text type,
such as a tale or a story. Rather, narrativity constitutes a pattern of
organisation of texts which can be found in any text concerned with action,
be it a fairy tale such as Cinderella (Courtes, 1993) or a cooking recipe
(Greimas, 1987). Since my purpose in this chapter is to show that narrative
structures can also be found in the structure of organisational activities, I will
Greimas' Semiotics and the Analysis of Organisational Action 131
devote this section to a brief presentation of the main components of

Greimas narrative grammar l8 •
1.1 The narrative schema
The general model of the organisation of narrative Greimas proposed is

known as the narrative schema (Greimas, 1987) or narrative program
(Groupe d'Entrevernes, 1984). The narrative schema first defines an
inventory of "actants" forming a basic set of relations (figure 1).
Sender Receiver
\
Subject Object
/
Helper Opponent
Figure 1. The actants in Greimas' narrative schema
In a typical James Bond story, for example, Bond (here the subject), is
asked by M (a sender) to recover a stolen microfilm (an object) and to give it
back to a receiver (the British Government, The West, etc.). On his way,
Bond encounters helpers (like Q) and opponents (like Dr. No). The same set
of roles is to be found in fairy tales where a Prince is asked by a King to
recover a kidnapped Princess and to bring her back to the Kingdom, and
where he gets a magical weapon from a good fairy (a helper) but has to face
an opponent. As Greimas puts it: "Actants are beings or things that
participate in processes in any form whatsoever, be it only a walk-on part
and in the most passive way." (Greimas and Courtes, 1982). This notion is
preferred by Greimas to that of actor because it is neutral as far as the
essence or the "ontology" of the participant in the action process is
concerned. The actants are thus defined from a strictly formal point of view,
by their relations to each other. In Greimas' view, any interactions between
the actants can be conceived, at a more abstract level, in terms of exchange
of material or symbolic objects. For example, the sender gives an order to
the subject, the helper gives him tools and the subject's mission is typically
18 For a detailed presentation of Greimas' narrative grammar, see Perron and Danesi (1993),
Perron and Collins (1989) or Greimas (1987).
to transfer an object from an opponent to a receiver. Greimas uses here the

notion of communication which he extends to the transfer and circulation of
objects between subjects throughout the narrative (Greimas, 1987).
The second most important aspect of Greimas' theory is his conception
of the basic pattern of narrativity, composed of the successive states and
transformations that go to make up a story. This pattern of transformation is
what makes the sequence a narrative according to Greimas. Four basic
phases or transformations form the narrative program, the building block of
a narrative: manipulation, competence, performance, and sanction.
In the manipulation phase, a subject receives and accepts a duty or
obligation from a sender (M gives James Bond the mission of finding a
stolen object). It is the process through which a sender "manipulates" a
subject to have him to do something. The process typically involves either
authority (as in a James Bond story) which Greimas describes as the transfer
of a "having to do", or persuasion described as the transfer of a "willing to
do."19 The basic relation involved here is the sender-subject relationship.
In order to act a subject must also have the necessary ability. In the
competence phase, the subject receives and gives proof of his or her ability
or expertise (Q gives Bond his most recently designed gadgets). This is the
phase where the relation between a subject and his helpers are most
apparent. The typical transformations involved here are the communications
of objects (tools, weapons, knowledge or information) that will enable a
subject to act. In his terminology, Greimas describes these objects as
pertaining to two categories: "knowing how to do" and "being able to do".
Next is the performance phase, where the subject transforms the
undesired states by obtaining the object of his or her quest, called the object
of value (for example, Bond kills the villain and recovers the stolen object,
say a microfilm). Here again the transformation accomplished can be
described in terms of the transfer of an object between two or more actants.
The performance typically represents the core transformation in any
narrative, the transformation for which the subject was previously prepared
in the manipulation and competence phases.
At the sanction phase, the subject receives an acknowledgment of his or
her success or failure by virtue of a given system of values (Bond is
rewarded). Typically, this phase involves the relationship between a receiver
and the subject responsible for the performance. In a typical fairy tale, for
example, the Prince is rewarded for his deed by being allowed to marry the
19 I here use the English translation of the French original terminology found in the English
version of Greimas and Courtes semiotic dictionary (Greimas and Courtes, 1982), even
though some of the terms may sound unusual in English. The original French terms for the
modalities are vouloir-faire (wanting to do), devoir-faire (having to do), pouvoir-faire
(being able to do) and savoir-faire (knowing how to do).
Princess for example. However, the more general meaning of sanction has to
do with the recognition, that even at a strictly cognitive level, a
transformation has occurred and that a new state has been produced. Figure 2
summarIes the basic construction of a narrative sequence according to
Greimas.
Manipulation -+ Competence -+ Performance -+ Sanction

Phase Transfer of the Transfer of the Doing: Transfer Transfer of
'having to do' 'knowing how of the object of recognition.
and/or 'wanting to do' and/or value Acknowledge the
to do' 'being able to new state.
do'
Relation sender/subject subject/helpers SUbject/opponent subject/receiver
Example M gives a Bond is provided Bond recovers The British
mission to with information, the stolen Prime Minister
Bond. weapons or tools microfilm from rewards Bond for
byQ. his opponent. his deed.
Figure 2. Greimas' model of a narrative program
To complete our presentation of Greimas' model a final distinction

between two kinds of objects must be presented here. As we have seen, the
four types of transformations forming the narrative program involve
transactions between actors who get to playa specific role in the schema of
actants. According to Greimas, these transactions can be described at a more
abstract level in terms of the communication or transfer of objects between
subjects. He introduces an important distinction, however, between two
kinds of objects: the objects of value and the modal objects also called
modalities. The object of value is the object of the performance, the object
"valued" throughout the narrative and which constitutes the object of the
subject's quest. However, the other phases of the program involve objects as
well. These objects are treated by Greimas as modal objects for they define
the motivation and ability of the subject to realize the performance. In other
words, the transformations involved in manipulation and competence phases
concern the relation that the subject has to the performance. This is why the
modalities are defined in Greimas' terminology as having to do, wanting to
do, knowing how to do, and being able to do. Modal objects are the means of
action.
Behind the simplistic construction provided in my James Bond story,
Greimas would argue that there lies a fundamental organisation of
transformations that will be found in any description of action if it is to be
meaningful. A description of action must not only state a transformation (the
recovery of a microfilm), but also the sources of the will or the obligation of
the subject to perform the act (called manipulation) and the ability he or she
has or does not have to do so (called competence). Even if James Bond's
program of action has a clear deed at its core (the recovery of an object), it
cannot be reduced to it since it involves a complex path through which the
recovery of the object will be made desirable and possible and where it will
be sanctioned.
1.2 From the analysis of text to organisational analysis.
Only a handful of students of organisations have considered Greimas

semiotics (Cooren, 2000; Taylor and Lerner, 1996). When it has been
applied to organisational studies, narrative analysis has been mostly
considered as an approach to the study of organisational discourse and
sensemaking processes (Weick, 1995). A number of scholars have argued
that the narrative organisation of discourse can be linked to the organisation
of experience and the interpretation of organisational reality (see Brown,
1986, 1987; Mageo, 1992; Fisher, 1985; Mumby, 1987; 1993; Strine and
Pacanowsky, 1985; Taylor and Lerner, 1996; and Weick and Browning,
1986).
However in recent years, a growing number of researchers have started to
look at organisational action and structures from a semiotic perspective
using Greimas as a heuristic model to understand the structure of action and
interaction and its implications for our conceptions of organisation. The
point of this body of work is to apply Greimas' notion of a narrative program
to the enactment of organisational activities and the structuring of
organisation and not strictly to the study of organisational discourse. Taylor
(1993) and Taylor and Van Every (1999), for example, argue that the
structure of exchange described by Greimas, and the narrative structure of
action that derives from it, is a crucial element of the emergence of
organisation. Starting from the analysis of speech acts and their
organisational dimensions, Co oren (2000) made a similar argument using
Greimas' semiotics as a basis of his model of the "organizing property of
communication". Following the idea that narrative analysis can be used to
the analysis of organisational action, I will devote the rest of this chapter to
an illustration of the way in which a semiotic perspective can shed a light on
the structuring of organisational activities. Let me now describe the
organisational episode I intent to use as an illustration of the way in which
Greimas' narrative grammar can also be conceived as a grammar of
organising.
2. THE ORGANISATIONAL CONTEXT
The case presented here is drawn from a vast field study I conducted on
the preparation, functioning, and consequences of the public discussion
process in a Canadian mid-sized municipal administration (for the purpose
of this paper, I will call it "The City"). By autumn of 1993, the political
leaders in office at The City had reached the middle of their third term of
office. To keep a promise to maintaining open relationships with their fellow
citizens, the political leaders asked the Communications Service and its chief
(let's call him "Jerry") to set up a public discussion process where the
citizens could voice their concerns directly to their elected officials on any
municipal question. This consultation and discussion process was called
"Operation Dialogue." The core of the operation was a series of six
discussion evenings between the public and their local elected officials, to be
held the following spring in each district.
Over a period of 11 months of interviews and observations, I gathered
data pertaining to the preparation of the consultation process as well as the
analyses of the evenings by various actors. I was able to follow the
preparation and analysis of the consultation during the six months preceding
the evenings and the four months following them. I will now tum to a closer
examination of the approach taken by Jerry and his acolytes in preparing the
meetings between the elected officials and the public. The preparation
process involved four main components: the conducting of a survey, the
formulation of an agenda, the creation of a slideshow and a publicity
campaign. Let me describe each of them briefly.
The preparation of the Operation Dialogue survey began with the
formulation of a questionnaire, an activity that required the Operation's
initiators to decide what they wanted to know. A part of the answer to this
question was given by the experience of the previous consultations and by
the survey's first general objective, to wit: "to measure the degree of
satisfaction of [members of the public ]20 with regard to City services and to
various aspects of its administration."21 At the outset, it was known that the
survey would include questions of the sort "What is your evaluation, for the
entire city, of the quality of the following services: ... »22 What followed was
a list of the 14 main municipal services. Apart from surveys conducted in
20 In quoted extracts, I have systematically replaced names or adjectives permitting the

identification of the site of my research. For example, the expression "Torontonians"
becomes "members of the public" or simply "the public."
21 Ville de [LaCite], Dialogue 1994. Rapport de sondage (Service des communications, Avril
1994) 4.
22 Ville de [LaCite], Dialogue 1994. Rapport de sondage, Annexe I: Sondage (Service des
communications, Avril 1994).
The City in 1986 and 1989, the choice and formulation of questions were
informed by the text of a similar survey conducted the previous year in a
neighboring city. The elaboration of the questionnaire gave rise to the
selection, formulation and inscription of several kinds of hypotheses and
questions during many exchanges and meetings held to determine the
subjects and issues which would be presented to the sample of members of
the public. The various service directors participated in discussions with the
Mayor or his executive assistant and above all with Jerry who was in charge
of the entire operation. As questions were discarded or retained, Jerry wrote
drafts of the questions that were subsequently reworked several times over.
On the whole, the survey results surprised almost everyone, elected
officials and administrators alike. The analysis of the results presented in the
report to the Communication Service maintained that the public
overwhelmingly supported the budget cuts (89.1% were for them), but
wished that service levels be maintained, even if this entailed a slight
increase in taxes (71.6% were in favor of this option). Satisfaction levels
with regard to municipal services largely surpassed the expected levels and
those obtained in the past. Several services received satisfaction levels
higher than 90% including recreation, culture, recycling collection, parks
and green space maintenance, cleanliness of public spaces, municipal
information. However, the City's internal report and the summary published
in the City newsletter explicitly designated four services for which
satisfaction levels were much lower. These included pet control (75.2%);
snow removal (74.3%); condition of the streets and sidewalks (72.2%); and
the garbage collection program with 71.5% of the respondents indicating
their approval of the program in place (respondents to this question could
only indicate their approval or disapproval of the program in place).
The next process involved formulating an agenda for the discussions to
be held with the citizens. Though intended to permit members of the public
to interrogate elected municipal officials about anything pertaining to
municipal affairs, the Dialogue evenings still followed an agenda established
in the weeks leading up to the event. The goal of this agenda was to order the
evening's activities and to group citizens' questions around certain
anticipated themes, for example problems related to road conditions. This
agenda proposed five discussion themes, four of which dealt with specific
municipal services: 1) street and sidewalk conditions and parking; 2) snow
removal; 3) garbage collection and the collection of material for recycling;
and 4) the cleanliness of public spaces and dog control. These four themes
were selected from among all municipal services because they were the ones
with which members of the public were most dissatisfied, based on survey
findings. To Jerry and his colleagues, these themes were unavoidable. The
experience of previous consultation process had revealed to them that these
themes encompassed the majority of the questions by, and expectations in

terms of municipal services of, the voting public. The fifth theme dealt with
the tax freeze policy. The elected officials were seeking to assess how far
they could go in reducing municipal services after almost three years of tax
freezes. For Jerry, who had prepared the agenda, and for most of the City
administrators and councilors I interviewed, there was no doubt that these
issues would be raised with or without an agenda. They constituted the
"citizens' concerns." In elaborating the agenda, the organizers of the
Dialogue evenings were attempting to accomplish two goals. They were
trying to foreground the concerns of the public as well as conducting an
ordered discussion of these concerns.
In order to encourage the discussion with the public and to obtain their
comments, the organizers also felt that the public needed to be "informed"
about current matters affecting The City before soliciting their opinions and
comments. At the time of the 1994 Operation Dialogue, the most pressing
problem in the minds of the elected officials was the City's budget policy.
The issue was clearly laid out in the agenda, but it was also presented in a
slideshow shown to the public at the beginning of each one of the discussion
evenings. Entitled Let's talk about The City - the theme of the entire
operation - the slideshow provided an overview of the budget cuts adopted
by the City in the previous years.
Once all the texts viewed as necessary to conducting the Dialogue
evenings had been prepared, it remained for the organizers to mobilize an
important actor in this story - the citizens. Again under the slogan Let's talk
about The City, the Communication Service had conducted a newspaper
publicity campaign for a month prior to the evenings. Large billboards were
erected throughout the city with a view to soliciting the participation of the
public. Moreover, each household received a mailed invitation to participate
in the evenings and had told the date, time, and place of the discussion
evening to be held in their district. Written for the most part in the
imperative, Operation Dialogue's "advertising" texts emphasized the theme
of open discussions and relied on the presumed desire of citizens to let
elected officials know about what dissatisfied them or about their "needs."
Inasmuch as a hundred or so people attended each of the Dialogue
evenings, the organizers of the operation felt that their advertising campaign
had worked. For Jerry, this participation signaled that in the eyes of its
citizenry, the City was a credible interlocutor, able to listen to and to take
into account their comments without which the discussion would not have
been worth attending. If the City was able to get the public out on an April
evening to "talk about the City", then it was because the public saw in the
invitation an implicit and effective commitment on the part of the City to
listen actively to them. In the view of various actors, the invitation and its
acceptance by members of the public contained a clear contractual

dimension.
3. THE GRAMMAR OF ORGANISING THE

OPERATION DIALOGUE
What organising grammar is operating in the Operation Dialogue? How

can we describe the organising of the resources, rules and scripts these actors
relied on in setting up the discussion evenings? In what way do the elements
of the Operation Dialogue constitute a narrative? To answer these questions,
we have to look at each element in its relationships to the interactions that
took place during the evenings. This perspective reveals the Operation
Dialogue as a narrative program.
3.1 Dialogue as a Performance
From its inception, the Operation Dialogue was already turning around
the interactions between the elected officials and members of the public
during the discussion evenings. The thinking about the operation by its
principal actors was oriented entirely towards the public's speech, which had
to be made possible and also had to be orchestrated. It was necessary to build
from scratch a speaking situation in which citizens could make their
concerns known to the City's elected officials. As such, this first general
framework for Operation Dialogue made the public's voice into a veritable
object of a quest or an object of value. The valorization of the public's
speech constituted a determining element of the meaning accorded to the
operation by the actors involved in it. This value effectively conferred a
narrative dimension on the whole operation.
Characteristic of narratives, the first actantial relationship was the
relationship of desire between the subject of the action, which I will refer to
the "City" and the object of value constituted by the "speech" or "voice" of
the public delivering their concerns and opinions. From this perspective, we
can consider the whole of Operation Dialogue as a program intended to
establish a relationship between the City and this object, to be acquired. This
relationship is represented schematically in Figure 3:
Subject
The City
Figure 3. Subject/Object in the Operation Dialogue

The value accorded to the interventions by the citizens, like that of any
object implicated in a narrative program, is, however, a function of the
relationships it has with other actants in a variety of different narrative
programs. Before noting how the Operation Dialogue came to be constituted
as a narrative program in which several actors were placed in relationship to
one another, we saw how the actors were initially involved in varied action
programs. When the Operation Dialogue was first undertaken, each of its
groups of actors had its own narrative, a local narrative in which the public's
interventions were not first and foremost. For example, as Jerry admitted, he
was looking beyond the operation to establish the legitimacy and usefulness
of his new department in the municipal apparatus. Councilor-members of the
governing party hoped that the comments expressed by the public would
persuade the Mayor to have the garbage collected a second time each week.
Elected members of the opposition party were just as interested as their
colleagues affiliated with the Mayor in learning the main sources of
discontent among the citizenry. Several administrators in the Technical
Services Department, believing the elected officials to be more sensitive to
the public's grievances than to their own, and hoped that the public's
comments would induce the elected officials to soften cuts to services. For
his part, the head of the finance department hoped that the operation would
give some life to his reform of the water billing system. Even the citizenry,
who had yet to become a part of the operation through the translation of their
interests by the administrators, would find a way into the Operation
Dialogue'S narrative program to transmit their views of The City affairs. The
value of the public'S speech as object of a quest was due as much to local
quests as to the Operation Dialogue's overall narrative. This value varies
depending on one's point of departure but culminated in the Dialogue
program itself.
The Operation Dialogue was progressively constituted as a common
program in which the City administrators and elected officials of all parties
became involved. It became a sort of obligatory narrative program. It is
important to note that this was not imposed on the actors in a "wanting to
do" modality through the profit-sharing work of a promoter of the operation,
but through the modality of "having to do" dictated by the City Council in
voting to hold the operation. Here, the City Council was an institutional
actor to which the distribution of modalities in the municipal organisation
conferred the ability to act- different from the other actors in the municipal
organisation. Engaged on their respective pathways, all the actors
nevertheless became involved in their own script within the Operation
Dialogue's narrative program.
3.2 The modalities of the Dialogue
The operation's narrative dimension is due, however, to more than the

relationship between a subject and an object of value. The narrative
organisation of the Operation Dialogue was also, and more importantly,
manifested in the intervention of the modalities of the discussion with the
citizenry. The dialogue of the discussion evenings, considered here as a
performance, was in fact heavily modalized by the elements of the Operation
Dialogue presented above, by the conclusions drawn from the survey and by
the agenda. I would like to argue that these elements intervened as modal
objects in the interactions comprising the Dialogue evenings and that in
doing so they constructed the motivation and the competence of both parties
in regard of the Dialogue. Let me review one by one the elements presented
above to disclose the modalities they introduced into the relationships the
elected officials and the citizenry had to the interactions during the
discussion evenings.
3.2.1 The Elected Officials' Narrative Program
When asked about the reasons behind the Dialogue evenings, the elected
officials invariably begin by invoking the promises made in this regard
during the preceding election campaign. Some even went so far as to point to
their party's platform with its clear statement that an operation of this sort
must be held midway through each term of office. The discussions with the
citizenry were thus initially conducted in a "having to do" mode, with
obligation arising from a prior electoral promise. However, this first
modalization is not a mere adaptation of the elected officials conduct to the
text of the platform an its script. In light of the elected officials concerns', it
appears clearly as a translation of both their current concerns' and the text of
the platform.
The elected officials did not limit the reasons for holding the discussions
to election promises. They stressed their own desire to learn about the
"citizens' concerns," to remain close to the voters, to "take the population's
pulse" so as to understand better their needs. Certain administrators went
further and saw the operation as a good way for elected officials to learn
how to face the citizenry during the next election campaign. Whatever the
case, the desire of the elected officials to understand the sources of
citizenry's dissatisfaction was beyond doubt. The relationship that the
elected officials had with the discussions therefore involved a "wanting to
do" mode. Promises made during the previous election campaign ("having to
do") and the desire to understand the concerns of the citizenry ("wanting to
do") form the essence of semiotic manipulation, that is, the elements pushing
a subject (the elected officials) to undertake an action program.
My description of the preparation of the Dialogue evenings also enables
me to identify the competence modalities informing the relationships the
elected officials had with the carrying out of the discussions. In this regard,
the survey conducted by the City played a central role inasmuch as it gave
the elected officials knowledge about the public's opinions, knowledge
which would be factored into the planning of the evenings, as we saw earlier.
The organisation of these evenings was also strongly influenced by the
experience of previous consultation sessions. These evenings provided
certain elected officials and the head of the communication service with
'know-how', and played a central role in planning the evenings especially
with regard to elaborating the agenda and in the logistical organisation ofthe
operation.
The relationships the elected officials and the City administrators had
with the holding of the Operation Dialogue discussions was also modalized
by a set of "being able to do" elements, as Greimas would categorize them.
Among the elements noted above, let me mention the agenda. The director
of the communications service viewed it as a way of facilitating the
interventions of all concerned. In addition to all the "being able to do"
elements held by the City, one must also include its ability to prepare and
conduct a survey, to reserve meeting halls, and so on. The preparation of the
evenings mobilized already distributed modal objects within the City
organisation or the community. Consequently, these modal objects were held
by the Communication Service. The Communication Service also influenced
by the construction of modal objects specific to the holding of the Dialogue
discussion evenings.
3.2.2 The Citizens' Narrative program
The preparation of the Operation Dialogue also gave rise to the setting in
place of a collection of modalities influencing the citizens' relationships with
the performance expected of them, and which consisted in attending the
event in order to tell their elected officials what there concerns were. The
repeated invitations by the Communication Service intended to get the
citizenry to participate in the operation constituted in this respect a clear
semiotic manipulation effort. With regard to the citizenry, the City was
engaged in an activity intended to get them to want to come and "talk about
the City".
This manipulation activity played a crucial role in the continuation of
Operation Dialogue. Among other things, it included, as is the case for any
kind of semiotic manipulation, the creation of a fiduciary contract linking the
initiator of the performance-the City in this case-to the subject of this same
performance-the citizenry, who were asked to come and let their concerns
be known. When accepted by a subject, this "executory formula" entails the
creation of a fiduciary contract linking the party requesting the action and
the subject of the action. We saw earlier that the individuals in the
Communication Service who were conducting this persuasive activity came
close themselves to examining their actions in the same way, inasmuch as
their reading of the situation, and in particular the response by the citizenry,
brought into play the notion of "credibility." They attached the highest
importance to this aspect of the relations with the citizenry, and pushed their
colleagues to deliver the goods. This crucial stage, as we shall see once again
below, set in place this "fiduciary contract" between the City and its citizens,
within the framework of which the actions of the citizens would be given a
meaning, even in the eyes of our devoted City administrators.
The manipulation of the subjects, that is, the citizens in Operation
Dialogue's narrative program, continued in the City's attempt to set the
agenda for the discussion sessions. Here, the citizens' interventions were
partially located in the "having to do" mode, inasmuch as certain issues were
indicated as being necessary objects of discussion, given the survey results.
The entire argument of the organizers of the evenings rested on the fact, as
we saw above, that this "having to do" they had imposed on the discussion
evenings would be assimilated into the "wanting to do" of the citizenry, a
link established by the survey results.
Lastly, the preparation of Operation Dialogue also had an influence on
the competence the citizens would need to mobilize in accomplishing the
performance "Let's talk about the City." This competence is first implicated
in the realization of the slideshow and its presentation at the beginning of
each evening. Indeed, we saw that the main objective of the slideshow was
to enable the public to make a contribution to the discussion, which would
be based on their understanding of the context in which the City and its
administration were operating. The preparation of the evenings thus
introduced an additional modality -knowing- in the public's relationship to
the discussion, thus building its competence to intervene in the discussion.
Finally, the sanction or evaluation of the performance appears clearly in
the analysis of the head of the Communications Services who maintained
that the level of the citizen's participation to the operation (over a hundred
citizens in each district) made the whole operation a success. The interviews
I conducted later on have shown that the elected officials, including the
mayor, share this evaluation of the process. Figure 4 summarizes the
narrative structure of organising involved in the setting up and the
conducting of the Operation Dialogue.
NARRATIVE PROGRAM: Gettin the citizens' voice
!MANIPULATION!
EO commit to maintain open relationships with the citizens (having to do) .
EO has interests in knowing the citizens' concerns (wanting to do).
ONWETENCE(SUBPROGR
!MANIPULATION!
EO mandate the CS to set up the discussion process (transfer of a
having to do).
CS accepts
f¢ONWETENcEj
CS mobilizes its resources and expertise to set up the process
(knowing and being able to do).
!PERFORMANCE 11
CS conducts the survey
EO get the results of the survey (knowing).
CS fonnulate a discussion agenda

EO get the agenda (being able to proceed).
!PERFORMANCE 2 (SUBPROGRAM)i
!MANIPULATION!
CS holds a publicity campaign to get the citizens' to
participate (transfer of a wanting to do).
About six hundred citizens accept (attend the
discussion evening).
iCONWETENCEj
CS prepares a slides how to provide the citizens with
relevant information about the City situatio n
(transfer of a being able to discuss some issues).
lPERFORMANC§
The citizens intervene and voice their concerns (transfer ofthe object of value)
EO listen and answer the citizens' questions
~ANCTION!
EO thanks the citizens for their participation.
EO and CS find Operation Dialogue a success,
NP: Narrative Program; EO: Elected official; CS: Communication Service
Figure 4. The narrative structure of the Operation Dialogue
From the analysis I have developed here, Operation Dialogue appears to

be structured like the most common story. Because of their prior
commitment and their political interests in being knowledgeable about the
citizens' concerns, the City'S officials undertake the "mission" of getting the
citizens to share their preoccupations directly with them. On their path to get
this transformation or performance to occur, they went through a number of
intermediary actions that shaped their competence and their capacity to

succeed in this endeavor. In that respect, both the conducting of the survey
and the formulation of the agenda can be considered as modal objects
building the elected officials' competence. The citizens are the other main
subject of this story since, after all, they are the ones from whom action is
expected. They are first the object of what Greimas would have called a
semiotic manipulation, where the City, here acting as a sender, engage in a
publicity campaign intended to get the public to participate or to persuade
the subject to engage in the proposed program. Then, once the citizens had
accepted to participate in attending the discussion evening, the public is
going through preliminary phases too, which will make them more
"competent" to intervene in the discussion. I have argued that the agenda and
more importantly the content of the slideshow figure as modal objects
building that competence. The discussions per se, in my view, constitute the
heart of the whole operation and thus can easily be interpreted as the main
transformation of this narrative, since the City gets literally here the object of
value ofthe whole story, that is, the citizens' interventions. The process ends
with an evaluation where both the Communication Service and the elected
officials find the operation successful.
4. DISCUSSION
Having shown the similarities between the basic structure of a narrative,

as described by Greimas, and the organisation of the Operation Dialogue, I
would like to extend my analysis to a discussion of the implications of this
semiotic model for the study of organisational action.
The first aspect of the narrative model relevant to organisational analysis
is its treatment of the problem of agency. The narrative model can in effect
be considered as a model of purposeful action (Cooren, 2000), since at the
heart of the very notion of narrative lies the relation of desire between a
subject and an object. Narratives are about goal-oriented transformations and
this orientation is the very basis of the meaning of action in this semiotic
perspective. But Greimas' model does not merely take that intentionally for
granted; it accounts for its very emergence through its analysis of the
modalities of manipulation. The notion of manipulation, interpreted as the
transfer of a having to do (through authority) or wanting to do (through
persuasion), displaces the locus and the source of agency from the actor to
the transactions the actor engages in. If one looks for the origin of an action,
according to this perspective, one will always fall back on transactions
where the apparent actor got his "intention" from various actors. So are the
citizens who voice their concerns after having being persuaded to do so by
the City and other senders absents in my account. So are Jerry and the
Communication Service, who go after the preoccupations of the public
because of the mandate they received from the City Council and the elected
officials. And so, again, are the elected officials themselves who were
motivated by others senders, among which were the citizens from whom
they received their office. As Latour puts it: "Indeed, action cannot have a
point of origin except at the price of stopping the circulation, or the series of
transformations whose movement continually traces the social body"
(Latour, 1996, p. 237). Agency still exists, it is just not where we thought it
was.
The treatment of agency in a reflection on the organisations is crucial
because, as most contemporary theorists argue, an explanation of the
structuration of organisations relies ultimately on an account of the relations
binding the idea of human agency to the undeniable action of structural
constraints (Giddens, 1984; Banks and Riley, 1993; Conrad, 1993). The
semiotic analysis of action can provide the basis of a new perspective on this
traditional dilemma, for in addition to incorporate a notion of agency, it also
offers a view on the sources and functioning of structures through the notion
of modality. Indeed, a great deal of what organisation theorists call
organisational structures can be understood in terms of the modal objects
shaping the relations between the actors and their action in a program of
action. Let me develop this point a bit further.
In elaborating his structuration theory, Giddens (1984) urged social
theorists to conceive of social structure as the medium of action as well as
the constraints imposed on it. In the dynamic of structuration, the
mobilization of structures in a course of action as a mean to achieve a
purpose is what maintain the very existence of these structures largely, in
Giddens' view, as an unintended consequence of action. Semiotic theory
gives us here an analytic language to achieve such a conceptualisation. The
institutional structure that gives the City Council its authority over the
administration cannot be more clearly described than it is in the transaction
where the Council mandates (tranfer a "having to do") the Communication
Service to set up the discussion process. The organisational hierarchy is
translated here in a hierarchy of narrative programs where the
Communication Service's program to set up the operation appears as a
subprogram or "sub-mission" (Cooren, 2000) embedded in the City Council
program of getting the voice of the citizens. Note that the same is true of any
James Bond narrative. Bond engages in a mission of recovering a stolen
secret. Q, for his part, is engaged in his own program, which is to provide
Bond with the tools and gadgets secret agents might need. The point is that
these two narratives do not only intersect, they are hierarchically organized,
in the same way as those of Jerry and the City Council are. Q's program is
actually a sub-program embedded in that of Bond, whose mission is in tum

embedded in that of M. and the British Government. Implicit, then, in
Greimas' semiotic account of action is the hierarchical structuration of
programs of action, a central feature of organisational structuring. This
structuration is implied in any organisational action, as Giddens speculated,
since all action engages modalities in the forms of modal objets, either
material or symbolic, mobilized by the actors. From this perspective, the
modalities of action, either those defining an actor's motivation (having to do
or wanting to do) or competence (knowing how to do or being able to do),
are actually what makes an action organisational. Since they define the
relationship the actor has to his own action (or his performance), the modal
objects literally incorporate the structural aspects of organisations. As
Greimas puts it: "the object appears to be a locus of fixation, a locus of
circumstantial clustering of the value determination." (Greimas, 1987, p. 86).
Structures, then, also exist in the semiotic perspective, but, like agency,
they are just not were we thought they were. They are not over the heads of
organisational actors, acting on them somehow as an invisible hand. They
are fixed on the objects engaged in the daily transactions of these actors, and
are reproduced through the very circulation of these objects and the
structuring that follows from it. In the same way, all the people, objects and
texts engaged in the Operation Dialogue's narrative program, by entering in
relationships with each other in this course of action, also shaped the very
organisation of the City.
Overall, then, I think a number of features of the semiotic narrative
analysis suggested by Greimas make the case that semiotic theory can
provide the field of organisational studies with a perspective and a language
which usefulness goes beyond the analysis of the issues of meaning and
interpretation in organisation. At the very least, the insights suggested by
semiotics on traditional problems of organisation theory deserve further
investigations.
5. CONCLUSION
Ricoeur (1986) once suggested that social theorists consider action as a

meaningful text. Brown (1987) went further to argue that semiotic analysis
should be considered as a way to account for the "superintenteded" or
deteremined aspects of social action within a context of assuming the
primacy of human authorship. But, as far as I know, neither of them engaged
into a systematic an empirical exploration of this intuition. My intention in
this chapter has been to make a first and very preliminary step in that
direction, in order to see the implications of such a conceptual move for a
few basic issues of organisational theory. I undertook this task using

Greimas'semiotics as a conceptual and methodological point of departure.
The empirical case for the semiotic analysis of organisational action I have
provided in this chapter reveal, I think, the potential strengths of this mode
of explanation in the study of organisation.
This is not to say, however, that Greimas' perspective is not to be
amended in the process. Regarding my argument that organisational
structures can be understood as the modalities of organisational activities, for
example, further organisational studies might probably lead to a revision of
Greimas' list of the four basic categories of modality. Besides knowing,
being able, having to and wanting, one might need to consider other
modalities of action in organisational life, if we are to account for the
unintended consequences of action (Giddens, 1984) or for the retrospective
character of the interpretation of action, another pervasive aspect of
organisational life (Weick, 1979; Perrow, 1986). Another feature of Greimas'
semiotics that should be reconsidered in the context of organisational studies
is the distinction between material objects, such as tools, and symbolic
objects, such as an authorization to do something. While Greimas' general
category of "being able to do" encompasses both of the latter, some
important distinctions at a lower level of abstraction might be highly
significant in order to understand the kinds of power, for example, involved
in action and the structuration of organisation. In sum, almost everything
remains to be done in the exploration of semiotic theory as a conceptual
resource for our understanding of the complexities of organisational action.
REFERENCES
Bantz, C. R. 1989. Organizing and The Social Psychology of Organizing. Communication
Studies, 40(4),231-240.
Banks, S. P. and Riley, P. 1993. Structuration theory as an ontology for communication
research. Communication Yearbook, 16, 167-195.
Berger, P. and Luckmann, T. 1966. The social construction of reality. Garden City, NY:
Doubleday.
Brown, M. H. 1986. Sense making and narrative forms: Reality construction in organizations.
In L. Thayer ed., Communication -> Organization: Emerging Perspectives J, Norwood,
N.J.: Ablex, 71-84.
Brown, R. H. 1987. Society as text: Essays on rhetoric, reason, and reality. Chicago: The
University of Chicago Press.
Conrad, C. 1993. Rhetorical/Communication theory as an ontology for structurational
research. Communication Yearbook, 16, 197-208.
Cooren, F. 2000. The organizing property of communication. Amsterdam: John Benjamins.
Courtes, 1. 1993. Semiotique narrative et discursive. Paris: Hachette.
Groupe d'Entrevernes. 1984. L'analyse semiotique des textes. Lyon, France: Presses
universitaires de Lyon.
Fisher, W. R. 1985. The narrative paradigm: An elaboration. Communication Monographs,
52(December),347-367.
Giddens, A 1984. The constitution of society. Berkeley: University of California Press.
Greimas, A. J. 1983. Structural semantics: an attempt at method (Daniele Mc Dowell, Ronald
Schleifer and Alan Velie, Trans.). Lincoln: University of Nebraska Press.
Greimas, A J. 1987. On meaning: Selected writings in semiotic theory (Paul 1. Perron and
Frank H. Collins, Trans.). Minneapolis, Minn.: University of Minnesota Press.
Greimas, A. J., and Courtes, 1. 1982. Semiotics and language: An analytical dictionary (Larry
Crist et ai, Trans.). Bloomington, IN: Indiana University Press.
Latour, B. 1996. On interobjectivity. Mind, Culture, and Activity, 3(4), 228-245.
Levi-Strauss, C. 1958. Anthropologie structurale. Paris: Pion.
Mageo, J. M. 1992. Submergeed forms: Properties of plot in narrative discourse. Semiotica,
92(1-2),49-73.
Mumby, D. K. 1987. The Political fonction of narrative in organizations. Communication
Monographs, 54 (June), 113-127.
Mumby, D. K. ed. 1993. Narrative and Social Control: Critical Perspectives. Newbury Park,
CA: Sage.
Perron, P., and Collins, F. eds. 1989. Paris school semiotics I: Theory. Amsterdam: John
Benjamins.
Perron, P., and Danesi, M. 1993. AJ. Greimas and narrative cognition. Toronto: Toronto
Semiotic CircielUniversity of Toronto Press.
Perrow, C. (1986). Complex organizations: A critical essay (3rd ed.). New York: Random
House.
Poole, M. S., and McPhee, R. D. 1983. A structurational analysis of organizational climate.
In: L. L. Putnam and M. E. Pacanowsky, Eds. Communication and organization: An
interpretive approach. Beverly Hills, CA: Sage, 195-219.
Propp, V. (1968). The morphology of the folktale. Bloomington, IN: Indiana University Press.
Putnam, L. L., and Pacanowsky, M. E. eds. 1983. Communication and organization: An
interpretive approach. Beverly Hills, CA: Sage.
Ranson, S., Hinings, B., and Greenwood, R. 1980. The structuring of organizational
structures. Administrative Science Quarterly, 25, 1-17.
Ricoeur, P. 1986. From text to action: Essays in hermeneutic II. Evanston, II.: Northwestern
University Press.
Robichaud, D. 1999. Interaction and textuality in organizing: Illustrations from a public
discussion process. The Communication Review, 4( 1-2), 103-124.
Strine, M. S., and Pacanowsky, M. E. 1985. How to read interpretive accounts of
organizational life: Narrative bases of textual authority. Southern Speech Communication
Journal, 50(3), 283-297.
Taylor, 1. R. 1993. Rethinking the theory of organizational communication. New York:
Ablex.
Taylor, 1. R., and Van Every, E. 1. 1999. The emergent organization. Communication as its
site and surface. Hillsdale, N.J.: Lawrence Erlbaum Ass.
Taylor, J. R., and Lerner, L. 1996. Making sense of sensmaking: How managers construct
their organization through their talk. Studies in Cultures, Organizations and Societies, 2,
257-286.
Weick, K. E. 1979. The social psychology of organizing. New York: Random House.
Weick, K. E. 1995. Sensemaking in organizations. Thousand Oaks, CA: Sage Publications.
Weick, K. E., and Browning, L. D. 1986. Argument and narration in organizational

Communication. Journal of Mange ment, 12(2),243-259.
11
ORGANISATIONAL SYSTEMS
Chapter 7
Reconsidering the Standard: a Semiotic Model of

Organisations
B. van Heusden and R.J. Joma
1. INTRODUCTION
Although in general we have little or no difficulty in speaking and

thinking about organisations, describing the empirical reality of
organisations is far from easy. Where should we look for it? How should we
study it? As is true for most cultural phenomena, organisations are markedly
elusive. They cannot be treated as empirical entities. What you perceive,
when 'looking' at organisations, are artefacts (buildings, machines) and
behaviour (linguistic and other). However, neither artefacts, nor behaviour
are 'organisation-like' in themselves. What is needed, therefore, is
something that relates artefacts and behaviour to create a more or less
coherent whole. Such a relation is a representation, shared, at least in part, by
a number of interacting actors. It is the representation that gives both
artefacts and behaviour their meaning. Clearly, we take representations
('symbols') not only as referring to something mental activities act upon,
but as a specific type of mental activity, also referred to as 'cognition'. If
anywhere, therefore, an organisation resides in the mental activities of
actors.
An organisation, from this point of view, is basically a representation that
structures the interaction among a group of people. It structures a system of
actors (a multi-actor system). This does not necessarily imply, however, that
154 B. van Heusden and R.J. Jorna
all the actors involved must represent an organisation in the same way. Nor
does it mean that actors represent an organisation in one way only. In fact,
actors often have to deal with conflicting representations. Take the case, for
instance, of a manager working for an international oil company, who also
sympathises with Greenpeace. As a representation, an organisation is
continuously subject to debate and change, and exists only insofar as the
interacting members of the group (re)-construct it time and again. An
organisation is not, when compared to the individuals involved, a different
level of aggregation (group versus individual). It must be conceived of as
belonging to a different ontological level. What people participating in an
organisation need is a shared representation of their interactive behaviour.
Such a shared representation is acquired through a complex semiotic
process.
We thus study organisations, understood, for the moment, in the largest
sense of the term, as systems of semiotic interactive behaviour, that is, as
multi-actor systems. These systems have a certain complexity. This
complexity is assumed to be a function of the number and the types of
behaviour involved (the quantitative and qualitative complexity,
respectively). More particularly, as we are dealing with representations, we
should focus on the semiotic dimension of organisational behaviour. Looked
at as a semiotic system (or set of semiotic systems), an organisation results
from an ongoing process of semiosis in individual actors. These individual
actors are situated in a context that is both physical and social. As a semiotic
reality, an organisation is an aspect of the knowledge available to and shared
by individual actors. Organisations are part of the semiotic (cognitive,
affective, volitional) structures individuals use to give form to their actions
and their environment. An organisation is part of the knowledge that certain
actors use to interpret data, that is, to construct information (cf. Boisot,
1995). At the same time, an organisation relies on specific knowledge for its
functioning (on technological knowledge, for instance, and on knowledge
about resources and markets, on social knowledge, as well as on general
knowledge about the world). A 'case for semiotics' is the description of the
organisational process of sense making (Choo, 1998). As innovation and
(knowledge) creation are semiotic processes, how do they relate to the
organisation as a semiotic entity? Models of organisational decision (cf.
Choo, 1998) can be read as types of representational strategy. In a similar
vein, the strategies adopted by organisations to achieve and manage
consensus must be understood as strategies used by individuals to achieve
Reconsidering the Standard: a Semiotic Model o/Organisations 155
and maintain common representations, that is, basically, as semiotic

strategies23 •
Semiotics is basic to the development of a theory of human action,
conceived as sense making, representation or semiosis (the three terms are
here considered to be synonymous). Meaning construction and knowledge
creation are both semiotic issues par excellence. The force of semiotics lies
in the theory of the sense making process (sense making, interpretation,
knowledge, information, and culture). Semiotics provides a (systematic) way
of looking at issues of meaning, interpretation and knowledge - a theory of
human culture. It offers a theory of the building blocks of culture: a theory of
signs and the process of representation. We may therefore expect, from an
organisational semiotics, a deeper insight into the specific aspects related to
organisational representation. It is all the more astonishing to find almost no
references to semiotic work in Weick's overview of literature on
organisational sense making (1995, 65-69).
As a semiotic entity, organisations structure the actions of the sign users.
The organisation exists as a semiotic tool, insofar as it structures the
environment and the actions of those who use it, that is, of those involved in
the organisation in one way or another. As a semiotic entity, an organisation
may need some restructuring now and then, to adapt it to changing
circumstances. Insofar as an organisation consists of the set of
representations that structure the actions of those involved, the knowledge
about the structure and the process of the (knowing) organisation is
knowledge of what is, in itself, a semiotic reality. As such it is meta-,
second-order, or semiotic knowledge. A semiotics of organisation deals with
the organisation, conceived as the representation of interpersonal relations
and actions. It will tell us about the functions of these representations, how
they are structured and what their particularities are. It may also be able to
predict how and when these representations will be innovative, thus leading
to an adaptation of the organisation to a changing environment, either
through changes within the organisation or through changes of the
organisation as a whole (single- and double-loop learning, respectively).
Thus we move from the structure of organisations to the organisation as a
representational construct, and from there to the structure of representations.
This shift of perspective involves a critique of the 'reification' of
organisations, of the 'reification' of culture in general (cf. Van Heusden and
lorna, 2000). The semiotic perspective definitely stands in the modernist,
23 Thus the three perspectives of organisational culture given in Choo (1998) can be related
directly to basic semiotic patterns. The integration-perspective may be characterised as
'metonymical', the differentiation-perspective as 'synecdochical', and the fragmentation-
perspective as 'ironical'.
symbolic tradition - taking organisations as a means of adaptation of

interacting individuals, and opposes both classical organic and mechanistic
models and post-modem laissez-faire approaches. It is thus reminiscent of
structuration theory (Giddens, 1984).
As (cognitive) mental activity, this representational behaviour is studied
by cognitive science. As representational activity, however, the cognitive
mental behaviour is the object of semiotics. Thus cognitive science and
semiotics meet in a joint effort to understand man, the animal symbolicum,
and human culture (cf. also Van Heusden and lorna, 1998; lorna and Van
Heusden, 2000).
2. METHODOLOGY: THE TWO APPROACHES IN

ORGANISATION THEORY
We face today, in organisation theory, an important challenge, which is

the development of a theoretical model of organisations, which could
function as an alternative to the mechanistic, goal-seeking framework. In
organisation theory (OT), as well as in information systems (IS), we meet
with two concurrent approaches: the 'hard' and the 'soft', respectively. The
dichotomy is also known as that of the 'functionalist' and 'computational' on
one side, versus the 'interpretive' and 'historical' on the other. The
opposition is often presented as one of 'logic' and 'structure' versus
'meaning', of observation and analysis versus understanding and
interpretation. Until now, it seems that the 'interpretive' approach has not
been able to provide us with a good model of culture as a sense-making
process. But on the other hand, the 'hard' approach hasn't been able to
provide us with a satisfying model of the interpretive dimension of human
culture. "The absence of an alternative model of organisation from the
interpretive school of thought is no doubt one reason why the model based
on rational-decision-making-in-pursuit-of-goals dominates in IS literature"
(Checkland & Holwell, 1998, 73). "There is no doubt that the goal-seeking
framework, leading to clear-cut analysis of levels of managerial creativity,
and hence well-defined types of information system, is exceptionally clear; it
is also intuitively convincing, as long as it is not questioned closely, and
eminently teachable. No wonder it gets taught!" (ibidem). "Viewed overall,
IS literature offers only the rather mechanistic model of organization which
derives from Simon and is based on rational decision making in pursuit of
organizational goals. The alternative interpretive strand of thinking is
currently the source of much lively work, but its writers do not yet offer a
detailed carefully-worked-out model of organization upon which are-think
of the IS field could be built" (ibidem, 74).
Reconsidering the Standard: a Semiotic Model of Organisations 157
Mutatis mutandis, the same can be said about economics and other social
sciences. On the one hand, Checkland & Howell are certainly right when
they say that meaning and interpretation have not been adequately theorised
or empirically investigated in research on organisations and information
systems. On the other hand, they are also creating a bogeyman that can
easily be put to fire. The position of Simon is much more sophisticated and
the distinction in hard and soft social science seems questionable. Simon has
been blamed for this framework because of his analysis of (rational) decision
making. This is only part of the story, however, and whether the attribution
in its totality holds is doubtful. Simon's work on cognition (1972) and his
analysis of the architecture of complexity and nearly decomposable elements
(1968) lead to a different appreciation (see also Simon, 1993). Instead we
suggest an open mind toward organisational issues taking into account the
fact that organisations consist of intelligent actors who by means of
(internalised) co-ordination mechanisms establish and accomplish
(conflicting) goals. Our goal in this paper is to suggest the possibility of
integrating the model of rational decision making in a model of sense
making (semiosis), in which rational decision making is only one of the
dimensions of a multi-dimensional semiotic process.
What we need, therefore, is a theory of the sense making process as a
multi-dimensional process. Theories being representations, such a theory of
representation is necessarily a second-order representation: it is itself a
representation of the representation process. As such, it must be able to
account for its own particular status as representational strategy, that is, for
its own representational 'regionality'. A systematic, rational theory of
representation thus provides us with a formal model of a process which, in
itself, is not necessarily formal.
Semiotics does not offer us, for the time being, a unified and generally
accepted theory of the semiotic process. We will therefore have to make an
attempt to bring together into our model the main semiotic approaches. First
the logical, focussing on the truth of representations; then the
phenomenological, which stresses the representation as constitutive of
human consciousness; also the linguistic-structuralistic approach,
emphasising the conventionality and codedness of representations; the
behaviourist, interested in signals more than in signs, the mathematical,
focussing on computation, and, finally, the evolutionary or genetic approach,
where representation is understood as a complex process of accommodation
and adaptation to the environment (for a more elaborated discussion, see
Van Heusden, 1997; 1999). Our semiotic model consists of basically two
parts: a difference or form problem generated by the dissonance between
memory and actuality and, secondly, a number of strategies used to deal with
this dissonance: images (concrete); concepts (conventions, codes, categories;
scripts); (logical) analysis (schema's; theories). Insofar as semiosis can be

reflexive, the model also comprises a second-order loop.
The starting point of the processes of representation is a dissonance
between memory and actuality, which results in form problems that are dealt
with through imagination, conceptualisation and analysis. A representation is
not a thing, but an activity, not an ergon, but energeia, in humboldtian terms.
It is (see above) an action in which a known form (stored in memory,
constituting memory!) is related to an actual situation. We may speak about
a relation of representation when the actual situation is recognised as (an
instance of) the known form. The known form becomes a sign of the actual
situation, which it represents. The process of representation entails a
continuous adaptation of memory to what is recognised in terms of memory
but always at the same time eludes memory to a certain point.
Representation is more than pattern-recognition (although it is that too); it is
basically the solving of pattern (or form)-problems.
Phenomenological semiotics stresses the experience of the representation
process, and the fact that it is constitutive of human consciousness -
consciousness seems to arise with the experience of difference. Moreover, it
stresses the concreteness of the experience, even the bodily sensations
(Merleau-Ponty; more recently Lakoff and Johnson, 1999). Linguistic
semiotics emphasises representation as conceptualisation - involving
conventions and codes (sets of conventions). Logical semiotics concentrates
on semiosis as logical analysis and argument, this being the only way to
secure knowledge. Behaviourism eliminates from its theory the properly
semiotic altogether and reduces semiosis to stimulus-response, the sign to a
signal. The genetic approach, finally, emphasises the complexity and the
dynamic nature of the process, the continuous adaptation and
accommodation. Moreover, it sees the different semiotic strategies as
cumulative phases in a process of phylogentic, ontogenetic, historical and
actual evolution.
We can now come back to the opposition between the 'hard' and the
'soft' approaches in organisation theory and information systems. It is clear
that the 'hard' approach deals with organisations (and with cultural reality in
general) in terms of logical analysis. Checkland's soft system methodology
focuses on the interpretation of images and concepts used in organisations to
make sense of data. The two approaches in OT and IS clearly do not exclude
each other - they represent different semiotic strategies in face of the
complex and problematic reality of organisations. It should not be too
difficult to integrate them into a genetic theory, which accounts for both of
them within the broader framework of a theory of organisation as a dynamic
semiotic process.
Reconsidering the Standard: a Semiotic Model ojOrganisations 159
The rational and the interpretative approaches can thus be related and
integrated within a broader semiotic framework. As a representation, an
organisation is not an objectively given entity, but has to be interpreted again
and again by the individuals working in and with it. This interpretative
process, as we know from a long hermeneutic tradition, often creates grave
problems and conflicts. Also, as a representation, an organisation may be
shared, discussed, negotiated and even enforced. And different individuals
for different goals may use the same representation. On the other hand,
however, the process of representation follows its strict semiotic logic and
can therefore be described and explained according to this logic.
3. SEMIOTIC STRATEGIES AND THE

STRUCTURE OF ORGANISATIONS
We will now elaborate upon this semiotic model, and show how the
reality of an organisation can be viewed, described, and explained in terms
of the semiotic model. With reference to organisations we will have to
emphasise relations with social systems theory. We will take Luhmann's
(1984) theory of self-organising and self-reproducing social systems as our
point of departure, taking the elements in the social system to be the
representing actors, and understanding the relations that constitute the
system as relations of representation (cf. also Berger and Luckmann, 1966).
What could make the process of organisational representation different from
other processes of representation? But maybe the question is wrongly put, as
it postulates a 'organisational representation'. How does the representation
process relate to organisations? In what sense are organisations the result of
representation processes and how does an organisation influence the process
of representation? Is there a reciprocal influence?
Taking the semiotic perspective as our point of departure, we can safely
assume that social organisation is an important prerequisite for the
development of conceptual, coded representations. These signs consist of the
unity of a signifier and a signified, and this unity is based upon conventions
(as in language, the Morse code or traffic signs). Conventions, however,
presuppose organisation. In tum, they make it easier to build and maintain
complex organisations. The two, therefore, strongly reinforce each other.
Concrete images and logical argument are much less dependent on social
organisation. The concrete image is primarily the result of the construction
of the perceptual apparatus, whereas the logical argument depends on the
structure or logic of reality. In all its complexity, the representation process
thus functions on the basis of the structure of the human organism, the social
organisation and the structure of reality. But for the time being we can state
that the coded sign, the unity of signifier and signified (or form and
meaning), is first and foremost a social fact. No codes without organisations.
In Weick's (1995) and Czarniawska-Joerges' (1992) terminology we are
here dealing with the 'generical subjective' dimension of representations. If
codes presuppose organisation, and if organisations presuppose codes, we
should ask ourselves whether organisations will emphasise coded
representations. Probably they will, but for the moment this is an open
question. What can be noticed, however, is the often routine-like behaviour
in organisations. To a certain degree, one could claim, an organisation is a
whole of routines: repetition is basic for its existence. This point of view,
that an organisation is basically a social semiotic fact, has been put forward
emphatically by the social philosopher Alfred Schutz (1967).
The lower threshold of this semiotic organisational behaviour is
determined by naturally determined behaviour. One can think about
emotions, reactions of fear, non-conscious responses, sense of threat, power-
hierarchy, etc. that is, general ethological phenomena.
Now how could the context or an organisation influence the semiotic
process of pattern-problem-solving? First of all by forcing upon its members
interpretation patterns, as well as pattern-problem-solving strategies. This is
done through learning, and socialisation. Secondly, by being an actuality that
requires representation and interpretation (how do we talk about the
organisation, how do we see it, how do we think about it). And thirdly, by
constituting an environment in which all sorts of representations are at stake,
emerging and changing due to the organisational interaction. The process
then becomes increasingly complex. In organisations an interaction develops
regarding representations (of goals, structure, mission and context of the
organisation) and this interaction is determined, at least in part, by the
second-order representation (or 'ideology') people have of themselves and
their interaction.
One can distinguish a restricted number of basic representations people
have of organisations. These derive from basic representational patterns.
They reflect the logic of representation. If the one-dimensional (concrete,
figurative) representation dominates, then actors will understand their
organisation as a concrete individual or object, and all the members are seen
as reflections of this single individual or object (the totem). This
organisation form is that of the clan. All members of the clan-like
organisation are basically equal, the only difference being their 'nearness' to
the totem (the result, for instance, of life experience). One could relate this to
what Wiley (1988) describes as the 'intersubjective' organisation. Actually,
one could argue that this is not 'really' an organisation, but rather a
collective subject!
When the two-dimensional, conceptual or coded representation

dominates, the organisation is understood as being a collective whole, with
an inside and an outside, and with parts that can have various functions, - as
in an organism - but always participate in the whole. The collective has a
clear hierarchy, it has vital parts (the head, for instance) as well as parts that
are dispensable (the limbs). Representations are not negotiable. They are
enforced. This corresponds to what Wiley calls' generic subjectivity' (1988).
The term is well chosen, as it reminds us of the concept of 'genre', which
refers to conventional patterns of representation. The relation between
individuals now has become categorical: they have become tokens of general
types. Subjects become conceptual entities, meanings in a common discourse
(which makes certain things, like downsizing, certainly easier). Again
according to Wiley, this is the 'real' level of organisation.
Barley (1986) stresses the importance of scripts for generic subjectivity.
Repetition is an important aspect of generic subjectivity. "Interactions that
attempt to manage uncertainty are a mixture of the intersubjective and the
generic subjective, which is something of a hallmark of organizational
sensemaking in general" (Weick 1995, 71). "I would argue that organizing
lies atop that movement between the intersubjective and the generically
subjective. By that I mean that organizing is a mixture of vivid, unique
intersubjective understandings that can be picked up, perpetuated, and
enlarged by people who did not participate in the original intersubjective
construction" (Weick 1995, 72).
When the logico-analytical representation dominates, we speak about a
market-like organisation. Again, as in the case of the clan, one could argue
that this is not, actually, an organisation at all - it is the organisation that
results from the absence of conventions, and that follows its 'necessary'
logic. The market is an open, but 'mechanistic' structure, and the
participants can fulfil different functions. Each participant can be replaced
and the organisation can be adapted at any time to serve a new goal. The
value of each element is determined by its function in view of the overall
goal. Professionalism and expertise are important factors. One could
eventually relate this type of organisation to the third level discerned by
Wiley, the level of the 'logic' of culture, of the extra-subjective. The
structure is determined, in the end, by 'objective' necessities, not by people.
It is no chance that Wiley sees both capitalism and mathematics as important
characteristics of this type of culture.
When the genetic model of the semiotic process dominates in the
representation of an organisation, this results in a complex, open and
therefore highly dynamic organisation. This organisation is in a sense a
mixture of the other types, but a reflexively self-conscious one, in which
people are aware of the semiotic nature of their organisation, which as such,
is subject to discussion and change. Scott (1987): "Open systems imagery

shifts attention from structure to process" (91). In such a complex
organisation the construction of meaning gets full attention, where other
organisations tend to preserve a once attained stable self-image, structure or
concept.
What we have tried to do is, in the end, to bring together a typology of
organisations and a theory of representation, conceived as a basic theory of
human action. The strategy of representation is reflected in the structure and
'behaviour' (metaphorically speaking, of course!) of organisations. Basic
patterns of representation work out in the structure of organisations. Which
is not surprising, as organisations are, in the end, the result of the
representational, that is semiotic, activity of humans. Depending on the
dominant semiotic (or representational) strategy, an individual will represent
an organisation and act accordingly. The culturally dominant semiotic
strategy may square with, or run against the individual strategy -
socialisation is therefore an important aspect of organisational sense making.
Each type of organisation creates its own type of subject, as is so beautifully
illustrated in Richard Sennett's The corrosion of character. The personal
consequences of work in the new capitalism (Sennett, 1998)24.
It might be interesting to relate our theoretical reflections to the research
into scripts. Is it possible to distinguish a relatively restricted number of
basic scripts? Or maybe one could distinguish fundamentally different ways
in which people deal with (various) scripts? We suppose both could be the
case (but see Barley, 1986). Eventually, one could relate the differences in a
relatively simple model, which could offer opportunities for further
formalisation.
4. CASTOR RESEARCH PROJECTS
In the context of the Castor-project, we have been investigating the

dynamics of organisational knowledge (knowledge change and knowledge
conversion). Our research is based on a semiotic framework (see Van
Heusden & lorna, 2000 for a more detailed expose, as well as the web-site of
the project: http://www.bdk.rug.nl/onderzoek/castor/). We are now in the
24 In a similar vein, Weick (1995) discerns three types of organisation: the rational and goal-
oriented organisation (Weber 1947; Simon 1957), the organic organisation (Roethlisberger
and Dickson 1939; Barnard 1938; Parsons 1960); and the open organisation (Buckley
1968; Boulding 1956; Katz and Kahn 1966). All three types are described in Scott (1987).
More recently, Henry Mintzberg and Ludo van der Heyden (Mintzberg and Van der
Heyden, 1999) gave a typology of organisations which also seems to derive from a, mostly
implicit, theory of representation.
phase of collecting data in different organisations. Thus we compare, for

instance, two organisational units in terms of types and forms of knowledge
the actors have, and in another case study, we study changes and conversions
of types of knowledge over time. Hypotheses that we would like to see
confirmed are:
- knowledge change always obeys the same logic (conversion from tacit to
theoretical knowledge without codification is impossible);
- conversion of knowledge from one knowledge type to another is possible
only when the organisation changes accordingly;
- direction and nature of innovation are different, depending on the type of
knowledge dominating in an organisation;
- we may distinguish three types of innovation: design, development,
discovery - design presupposes tacit knowledge; development
presupposes coded knowledge; discovery presupposes theoretical
knowledge;
- the diffusion (or distribution) of knowledge is dependent upon the way in
which knowledge is represented: not all knowledge can be represented and
distributed in the same way: tacit knowledge is distributed only via direct
contact, coded knowledge relies on the knowledge of codes; theoretical
knowledge requires abstraction (generally acquired through education);
- The storage of knowledge is also dependent upon the way in which
knowledge is represented: tacit knowledge is stored in individuals; coded
knowledge is stored in rules; theoretical knowledge is stored through
education.
Alexander van der Voort is doing research in the Royal Netherlands
Army. He applies the Castor knowledge-type model to explain differences in
the process of innovation of the management of information provision in the
Netherlands Army between the 'green' and the 'white' staff, respectively.
The 'greens' are those who, within the Dutch-German Army Corps, are
responsible for the management of the information provision of this Corps.
The Corps is the operational part of the army, deployed in war or
peacekeeping operations. The 'whites' are the personnel of the 'National
Commando', responsible for the management of the information provision,
supporting all the not directly military operational tasks of the army.
Although there is no a priori reason why different parts of the army should
have a different adoption rate of process innovation, in practice they clearly
do.
Mapping the knowledge situation of both 'greens' and 'whites', this
difference in organisational dynamics can be explained in terms of the
semiotic model. If the model is adequate it must be possible to give advice
regarding changes that could be made in the knowledge situation to improve
the innovation process. At the moment the research is in the data gathering
164 B. van Heusden and R.J Jorna
phase, using a measurement instrument that has been developed by the

Castor project members (Van der Voort, 2000).
Larissa Sjarbaini's research focuses on the conversion of knowledge in
the field of planning. Take, for instance, the situation of a scheduler in a
hospital scheduling the personnel of one unit. The scheduler does the
scheduling without the aid of a computer. Then the situation changes.
Instead of making a schedule without computer aid, the scheduler now starts
to make the schedules with the aid of planning support software. In using
this planning support software the scheduler realises that he was working
with more knowledge than he was aware of (this being is the assumption).
For example, after a night shift a nurse is not to be scheduled for a certain
amount of shifts. The scheduler used to automatically make a schedule in
which this criterion was met. But when he started working with the planning
support software he didn't realise that he used this knowledge to make a
schedule. So he forgot to put this information into the software program,
which resulted in a schedule that was not optimal. Running into this problem
with the planning support software has resulted in some knowledge
conversion for this scheduler.
In order to show whether a conversion of knowledge is actually taking
place in the above-mentioned example, the knowledge types of the scheduler
will be measured at three points in time. First, just before the scheduler starts
using the planning support software, a second time just after the scheduler
has started to use the planning support software and a third time about three
months after the scheduler has started to use the planning support software.
The knowledge of the schedulers will be measured with a questionnaire in
which questions are asked that concern the knowledge types a scheduler uses
in order to perform his task of scheduling. The research will be carried out
with two groups of schedulers. One group in an innovative situation as
described above, and one group of schedulers which is not. The non-
innovative scheduler group will consist of schedulers who are planning
without any computer aid and who continue to do so.
5. CONCLUSIONS
The (semiotic) theory of representation allows for a complex and

sophisticated approach in organisation theory, integrating seemingly
conflicting perspectives (the rational, the interpretative). It provides us with
a general and systematic model of representation, in which the rational
approach is itself explained as one possible semiotic strategy which
dominates in contemporary organisational theory and culture.
Organisations result from the interaction, in a multi-actor system, on the

basis of shared representations. The reality of organisations is the reality of a
complex semiotic process involving all interacting participants; a change of
organisation necessarily involves a change in the representations of the
actors and representations may vary considerably among actors within the
system. The structure of the representation process underlies organisation
structures; and we can thus relate organisation theory to a general theory of
representation and culture.
REFERENCES:
Barley, S., 1986, Technology as an occasion for structuring evidence from observations of
CAT scanners and the social order of radiology departments. In: Administrative Science
Quarterly 31,78-108.
Barnard, C.I., 1938, The functions of the executive. Cambridge, MA: Harvard UP.
Berger P.L., and Luckmann, Th., 1966, The social construction of reality: a treatise in the
sociology of knowledge. New York: Doubleday.
Boisot, Max H., 1995, Information Space. A framework for learning in organizations and
culture. London: Routledge.
Boulding, K.E., 1956, General systems theory: The skeleton of science. In: Management
Science 2, 197-208.
Buckley, W., 1968, Society as a complex adoptive system. In: Buckley, W., ed., Modem
systems research for the behavioral scientist. Chicago: Aldine, 490-513.
Checkland, P. and Holwell, S., 1998, Information, systems, and information systems: making
sense of the field. Chichester: Wiley.
Choo, C. W., 1998, The knowing organization. How organizations use information to
construct meaning, create knowledge, and make decisions. New York etc.: Oxford UP.
Czarniawska-Joerges, B., 1992, Exploring complex organizations: A cultural perspective.
Newbury Park, CA: Sage.
Giddens, A., 1984, The constitution of society. Berkeley: University of California Press.
Hatch, M.J., 1997, Organization theory. Modem, symbolic and postmodern perspectives.
Oxford: Oxford UP.
Heusden, B. van, 1997, Why literature? An inquiry into the nature of literary semiosis.
Tiibingen: Stauffenburg Verlag.
Heusden, B. van, 1999, The emergence of difference: some notes on the evolution of human
semiosis. In: Semiotica 127 1-4, 631-646.
Heusden, B. van, and Jorna, R., 1998, Toward organizational semiotics. In: Semiotic Review
of Books 10 1,
Heusden, B. van, and Jorna, R., 2000, Toward a semiotic theory of cognitive dynamics in
organizations. In: Organizational Semiotics. Berlin: Kluwer.
Jorna, R. and Heusden B. van, 2000, Cognitive dynamics: a framework to handle types of
knowledge. In: 1. Schreinemakers, ed., Knowledge management: the 7th ISMICK
conference. Rotterdam University Press, 128-145.
Katz, D. and Kahn, R.L., 1966, The social psychology of organizations. New York: John
Wiley.
166 B. van Heusden and RJ Jorna
Lakoff, G. & Johnson M., 1999, Philosophy in the flesh. The embodied mind and its
challenge to western thought. New York: Basic Books.
Luhmann, N., 1984, Social systems. Translated by J. Bednarz Jr. with D. Baecker. Stanford,
California: Stanford UP, 1995.
Mintzberg, H., and Van der Heyden, L., 1999, Organigraphs: drawing how companies really
work. In: Harvard Business Review, September-October, 87-94.
Newell, A., Shaw J.C., and Simon H.A., 1958, Elements of a Theory of Human Problem
Solving. In: Psychological review, 151.
Parsons, T., 1960, Structure and process in modern societies. Glencoe, IL: Free Press.
Roethlisberger, F.J. and Dickson, W.J., 1939, Management and the worker. Cambridge, MA:
Harvard UP.
Schutz, A., 1967, The phenomenology of the social world. Evanston, IL: Northwestern UP.
Scott, W.R., 1987, Organizations: rational, natural and open systems. Englewood Cliffs:
Prentice Hall.
Sennett, R., 1998, The corrosion of character. The personal consequences of work in the new
capitalism. New York & London: Norton.
Simon, H.A., 1957, A behavioral model of rational choice. In Simon, H.A., Models of man.
New York: John Wiley.
Simon, H.A., 1969, The sciences of the artificial. Cambridge, MA: The MIT Press.
Simon, H.A., 1993, Strategy and organizational evolution. In: Journal of strategic
management 14, Special Issue Winter 1993, 131-142.
Voort, A.G. van der, ed., 2000, Castor Werkbladen, versie 1.0 in Dutch, Published on
Internet: http://www.bdk.rug.nl/onderzoek/castor/
Weber, M., 1947, The theory of social and economic organization. Translated and edited by
A.H. Henderson & T. Parsons. Glencoe, IL: Free Press.
Weick, K.E., 1995, Sensemaking in organizations. Thousand Oaks, CA: Sage.
Wiley, N., 1988, The micro-macro problem in social theory. In: Sociological Theory 6, 254-
261.
Chapter 8
Using Semiotic Framework to Study Social

Consequences of IT
Michael S. H. Heng
1. INTRODUCTION
The functioning of infonnation systems depends upon physical resources,

human resources and societal resources. Physical resources refer to natural
resources, technologies and physical infrastructure. Human resources refer to
human labour, knowledge and skill in the broad sense of the words. Societal
resources refer to resources residing in moral and ethical system, institution,
culture, language and community. Examples of works incorporating such
ideas are Kling's (1987) Web model and the notion of infonnation systems
as social systems by Hirschheim (1989).
While the functioning of infonnation systems depends on physical,
human and social resources, the use of infonnation systems can also impact
on those resources. The literature documenting various aspects of this
phenomenon includes Dunlop and Kling 1991, Kling 1996. However there is
no framework with a strong theoretical underpinning to organize such
findings. While teaching infonnation systems and the social consequences of
IT, the author felt that the students needed a simple-to-understand
framework to provide some degree of coherence and structure for them.
Given its broad social orientation, I looked within semiotics for a tool to
help students attending a course on the impact and implications of IT on
economy and society to articulate their understanding of the social
168 Michael S. H Heng
consequences ofIT. This paper first investigates the possibilities of using the
organizational semiotic framework to structure our discourse then it reports
on introducing the framework to a group of master-level students ..
There are several important precursors to organisational semiotics. One
of the earliest in the field was the theory of information systems developed
in Sweden by Borje Langefors (1973) which his team incorporated into the
ISAC methodology. Another approach that developed shortly after is that of
Stamper(1973). Since then, semiotics has figured increasingly in information
systems research. Andersen in his Computer Semiotics (1991) united
linguistic and computing theories for the analysis, assessment and
evaluation. Clarke (2000) has shown how to apply the social semiotic and
functional linguistic theories of the Halliday School to information systems
problems.
For the study that I am reporting here, I use Stamper's semiotic
framework for two reasons. His semiotic 'ladder' belongs within MEASUR,
a wide pan-semiotic approach that attempts to provide a coherent picture
across all aspects of the analysis, design and realisation of organisations with
and without technology. Secondly, his work is quite well known within the
Dutch IS community, having served as an IS professor at a Dutch university.
There is sufficient reference literature in the Dutch university libraries
describing his framework and its application in IS development.
This framework has proved to be very useful in developing and using IT-
based information systems as well as studying IS-related activities such as
business process redesign and electronic commerce (Liu 2000; Heng and
Newman 2001; Stamper 2001; Chong and Liu 2001; Filipe, Liu and Sharp
2001; Barjis, Dietz and Liu 2001). As an exercise to investigate its potential,
I have focused on the three semiotics levels of semantics, pragmatics and
social world for examining the framework in this new domain of application.
We are asking whether a conceptual framework or theory with a proven
record in one area can be extended to other areas.
Using Semiotic Framework to Study Social Consequences of IT 169
ISOCIAL WORLD - beliefs, expectations,

Human Information commitments, contracts, law, culture, ...
Functions I;RAGM~TICS - inten~ions, communication,
conversatIons, negotIatIOns, ...
ISEMANTICS - meanings, propositions,
................................. validity, truth, signification, denotations, ...
The IT I~YNTACTICS - formal structure, language, logic,
Platform Data, records, deduction, software, files, ...
I~MPIRICS - pattern, variety, noise, entropy,
channel capacity, redundancy, efficiency, codes, ...
PHYSICAL WORLD - signals, traces, physical distinctions,
hardware, component density, speed, economics, ...
Figure 1. The semiotics framework (Stamper 1973, 1994)
Information technology (IT), as the term implies, refers to technology

that supports information related activities. From this perspective, various
forms of IT have played significant roles in our social, economic, and
cultural evolution for at least five thousand years, going back to the dawn of
writing, the invention of the abacus, the advent of talking drums, printing,
the Jacquard loom, etc. At the core of modem IT we now have the computer.
By attempting to match the changes resulting from earlier forms of IT to the
semiotic framework, can we see more clearly the potential for change
inherent in the most modem IT?
In this exercise, I chose writing, which is seldom mentioned when talking
of IT. In the light of human development, speech, agriculture, writing,
printing, telephone and radio have brought momentous changes. Speech
must have been the most important (Dampier 1961) but writing marks the
birth of history and human civilization as we know it.
The paper now reviews the impacts of writing and modem IT at the
levels of semantics, pragmatics and social world. After discussing the
findings and suggesting some future research, we report the students' use of
the framework to understand the consequences of IT.
2. SEMANTIC LEVEL
This level is preoccupied with issues related to meanings, propositions,

validity, truth, signification and denotation. The use of IT can affect our
perception of what has been traditionally accepted as truth, facts or evidence.
Writing has produced dramatic effects when introduced to an oral
culture. In her autobiography Out of Africa, Isak Dinesen (1954) describes
the response of the Kikuyu tribesmen to their first exposures to the written
texts. The effect of a piece of news was many times magnified when
imparted by writing. The messages that would have been received with
doubt and scorn if they had been given by word of mouth were now taken as
gospel truth. This has certain parallel in research methodology. It has been
observed that quantitative research is respectable in management research,
and this respectability is reinforced by the use of computer (Sanders 1982:
357). Video camera acts as a more reliable and objective "eye-witness" in
sports, thereby eroding some of the judgmental power of the referees. These
examples suggest that a piece of information acquires more credence and
respectability if a more advanced technology is used to represent, process
and transmit it. But this may be dangerous. The digitisation of images opens
up the possibility of manipulating the images, producing such photographs
as Marilyn Monroe dancing with Abraham Lincoln (Mitchell 1994). This
new technical advance underlines the importance of eyewitnesses in court
trials.
A crucial factor in the use of any IT in organizational life may be its
influence on people's perceptions. Semiotics reveals that our use of signs
helps to determine what we regard as reality. We now acknowledge the
social construction of reality in the social world. The field work of Luria
(1976) in the 1930s showed how people in the oral culture of Uzbekistan
quickly acquired abstract concepts, previously unknown to them when
introduced to writing. Millennia ago, writing transformed the spiritual world
by making possible the sacred texts that confer a kind of 'concrete reality'
upon some theological concepts that lie still at the roots of inter-communal
strife. The word is sometimes more crucial than reality. This is especially so
in political life. We must be careful when the use of modern IT in the world
of signs "dissolves" the boundary between what is real and what is
simulated, a potential danger in the misuse of AI. Virtual reality, so useful in
the world of engineering, must be treated with caution. We have already
seen that the popular view of history presented by Hollywood has displaced
the facts, as historians know them. This phenomenon has to be taken
seriously.
3. PRAGMATIC LEVEL
This level focuses on issues related to intentions, communication,

conversation and negotiation.
To oral culture, writing can revolutionise communications between
people. "Its essential service is to objectify speech, to provide language with
material correlative a set of visible signs. In this material form speech can be
transmitted over space and preserved over time; what people say and think
can be rescued from the transitory limits of oral communication." (Goody
1968: 1) Our unaided, oral speech acts express our intentions to people in the
immediate locality. Writing allows us to express intentions over great
distances and long periods of time. In his What Happened in History, the
historian Gordon Childe (1942) points this out emphasizing that sender and
receiver use the same conventional symbolism. The cumulative function of
writing in the preservation and dissemination of existing knowledge had
immense consequences for our civilization. Knowledge, once discovered,
need not be discovered again.
In other ways, writing expands our intellectual horizons and facilitates
the growth and accumulation of knowledge. It gave rise to new disciplines.
This would not have been possible without writing. Etymology as the study
of the origin and history of words is a tool used by social scientists and
philosophers in textual criticism. Logic was impossible when the facts
asserted orally stopped there, whereas, in a textual form, their logical
implications can be found. The setting down of speech enables us to separate
words, to manipulate their order and to develop forms of symbolic deductive
reasoning. Analytic studies of procedures, such as Aristotle's Art of Rhetoric,
depend for their existence on writing; they are produced not by the unaided
human mind but by the mind making use of a technology that has been
deeply incorporated into mental processes themselves. Writing also
influences our poetic style. Homer's poems are full of repetitive structures
that the storytellers and singers of the oral tradition who based their art on
memory.
Writing profoundly affected many human activities. The scope of
administrative and military control by the political centre was widened,
making it possible the emergence of large political unit such as the Roman
Empire (Innis 1972). Diplomatic exchanges between two states could avoid
the uncertainty associated with verbal messages; treaties could be encoded in
writing and signed.
For the expression of intentions, we attach more weight to a written
statement than a verbal one. "The signature effectively becomes a substitute
for the person, at least at the bottom of cheques. But it is not only a card of
identity, as individual as the print of the figure or the hand, but also an
assertion of truth or of consent. (Goody 1986: 152)"
4. SOCIAL WORLD LEVEL
This level refers to the products of information such as beliefs,

expectations, commitments, contracts, law and culture. These issues all
relate to the attitudes, norms and rules that socially govern the ways in which
people are disposed to act.
172 Michael S. H. Heng
Writing gave rise to law as we now know it. Havelock (1978) relates how
the concept of Platonic justice developed under the influence of writing out
of archaic evaluative accounts of human operations (oral 'situational
thinking) innocent of the concept of justice as such. In another independent
study, Goody (1986) considers at length how writing influences the concept
of law, its relation to legal rationality, the procedures, the institutions and
concept of law. He notes that in Europe the distinction between law and
custom is ultimately based on what was written before proclaiming it as law.
It implies that in societies without writing, even when courts exist, there is
no effective distinction between law and custom as sources of judicial
decision. Written records of court decision provide the material basis for
recourse to precedent in legal reasoning. Without it, each court sitting
represents a fresh start, for there is no permanent authority outside the court
to decide about the rules.
From the point of belief or ideology, writing altered people's perception
of the relation between past, present and future. Societies that possess
writing have a record of past events and are aware that they have a history.
Understanding history can serve to develop a sense of overall movement or
line of development that a society is following and, through conscious
political action, promote societal development in a certain direction
(Giddens 1993).
Writing, as the enabling technology that made law and history possible,
contributed, when mechanised by printing, to the making of nation states.
Each developed a distinct system of behavioural norms (laws) and cognitive
norms (history) defining itself. These sharp boundaries between
communities have had some unfortunate consequences.
Where is the computer, as an enabling technology, taking us at the social
level of analysis? An obvious answer appears to be: towards globalisation.
Business can operate globally as never before. International corporations
now compete in scale and power with the majority of nation states. The
norms governing these developments are being taken out of the nations into
institutions such as the IMF, World Bank, WTO. In terms of Gross National
Product the results are good but others argue that GNP treats all economic
activity, whether constructive or destructive, as equally valuable and thus
dispute the evaluative norms of classical economics (Daly and Cobb 1989).
However, these effects interact with those of other technologies. Television
and cinema beam images exhibiting the prosperity and mores of the OECD
nations across the world. These visual messages are highly motivating but
they do not help the viewers to act on their feelings. The global networks of
computers do. They provide a capacity to organise in response to the trends
in commercial globalisation. The web is full of discussion groups and
burgeoning pressure groups analysing and making ready to counter the
Using Semiotic Framework to Study Social Consequences ofIT 173
undesirable effects of economic globalisation. This rational trend in a most

desirable way is not only challenging established orthodoxies but is
dissolving old community boundaries and making others. But rational
developments appear not to tell the whole story. The incidents of 11
September would have been impossible to perpetrate without our new
technologies to support their funding and organisation.
5. DISCUSSION
Clearly we need to improve our foresight so that we may anticipate the

possible results of IT developments by actions to ward off their bad effects.
If the simple semiotic framework can alert us to the consequences of IT then
it is worth making it more widely known.
System builders must think beyond the well-understood issues of
hardware, signalling and software, or, their creations, like Frankenstein's
monster, may be difficult to control. Although Albert Einstein advised us to
approach our work with prudence and a deep sense of social responsibility,
we have atomic weapons and still have to learn how to control them.
Those attempting to divine the positive and negative implications of the
Internet and other forms of IT for our social life include social scientists such
as Castells (1996, 1997). We should listen to novelists such as Huxley
(Brave New World) and Orwell (Nineteen Eighty-four). Our enthusiasm for
rational solutions such as those driving global economic developments
should be tempered by Voltaire's satire of Leibniz, the father of all AI
representation languages, who rationally deduced that this is "the best of all
possible worlds". Douglass North, whose work earned him the Nobel Prize
in economics, pointed out that the financial 'bottom line' may not be the
right indicator of what is best. He said: "Strong moral and ethical codes of a
society is the cement of social stability which makes an economic system
viable. (North 1981: 47)" Thereby he challenged the designers of
information systems to find ways of signalling those moral and ethical
values to which financial information is supremely indifferent.
The effects on societal resources are rarely treated in information systems
literature. Non-financial values are embodied in our culture and community
institutions these are being recognised even in management literature, e.g. by
Nahapiet and Ghoshal (1998), quoting with approval Putnam's works (1993
and 1999) on social capital. Social capital is the product of our use of
information (see Stamper 2001, p. 117). Whether we augment or diminish
that capital depends to a large measure on how we use each exciting new IT.
174 Michael S. H. Heng
6. HOW A GROUP OF STUDENTS RECEIVED THE

MODEL
A group of students were asked to assess the value of Stamper's model as

a framework to understand the social consequences of IT. The students were
from Austria, Belgium, France, Italy, Mexico, The Netherlands, Spain,
Sweden and Switzerland - mostly international exchange students at the
Vrije Universiteit Amsterdam. They were generally in the final stage of their
MBA studies or equivalent, and were very motivated. They were attending a
workshop on the impact of IT on economy and society in spring 2001. The
course material consisted of book chapters and journal papers dealing with
the social aspects of use of IT.
I described Stamper's model and asked the students to read more about it
in the libraries. Three weeks later, they all answered a questionnaire
evaluating the model (see appendix 1). Of the fourteen, thirteen found the
model easy to understand. (Given their short acquaintance with the model,
minor wrong interpretations were permitted, as revealed in a sample in
appendix 2.)
Students were able to use the models to organize their understanding of
the social consequences of IT. I would not pretend to be able to organize all
the impacts of writing, let alone IT in general, within the semiotic
framework. But it is a promising candidate for structuring such impacts of
IT. The merit of the model is widely known when applied to the
development and use of IS. What can we say now is that these students were
helped in documenting, analysing and structuring the consequences of IT on
physical, human and societal resources.
REFERENCES
Andersen, P B 1991. A Semiotic Approach to Construction and Assessment of Computer
Systems, in Nissen, H. E. et al eds., 1991. Infonnation Systems Research, Amsterdam:
North Holland
Barjis, J.; Dietz, J. L. G. and K. Liu, 2001. Combining the DEMO Methodology with
Semiotic Methods in Business Process Modeling, in Liu, K.; Clarke, R. J.; Andersen, P. B.
and R. K. Stamper eds., 2001. Infonnation, Organisation and Technology: Studies in
Organisational Semiotics BostonlDordrechtILondon: Kluwer Academic Publishers,
Chapter 8, pp. 213-246
Castells, M 1996. The rise of network society. Oxford: Blackwell
Castells, M 1997. The power of identity. Oxford: Blackwell
Childe, G. 1942. What Happened in History, reprinted 1965. Hannondsworth, u.K.: Penguin
Chong, S. and K. Liu (200 I). A Semiotic Approach for Distinguishing Responsibilities in
Agent-Based Systems, in Liu, K.; Clarke, R. J.; Andersen, P. B. and R. K. Stamper eds.,
2001. Infonnation, Organisation and Technology: Studies in Organisational Semiotics

BostonlDordrechtILondon: Kluwer Academic Publishers, Chapter 6, pp. 173-186
Clarke, R. J. 2000. An Infonnation System in its Organisational Contexts: A Systemic
Semiotic Longitudinal Case Study, PhD Dissertation, Department ofInfonnation Systems,
University ofWollongong, Australia
Daly, H E and J B Cobb, 1989, For the Common Good: redirecting the economy towards
community, the environment and a sustainable future, London, Greenprint
Dampier, W C 1961. A history of science, 4th edition. Cambridge: Cambridge University
Press
Dinesen, I. 1954. Out of Africa and Shadows on the Grass. New York: Penguin
Dijksterhuis, E.J. 1961. The Mechanization of the World Picture. Oxford: Oxford University
Press
Dunlop, C. and Kling, Reds. 1991. Computerization and controversy: value conflicts and
social choices, I st edition. London: Academic Press
Filipe, J.; Liu, K. and B. Sharp, 200 I, A Semiotic Approach to Organisational Role Modelling
for Intelligent Agents, in Liu, K.; Clarke, R. J.; Andersen, P. B. and R. K. Stamper eds.
2001, Infonnation, Organisation and Technology: Studies in Organisational Semiotics
BostonlDordrechtILondon: Kluwer Academic Publishers, Chapter 7, pp. 187-211
Giddens, A. 1993. Sociology, 2nd edition. Cambridge: Polity Press:
Goody,J (ed.) 1968. Literacy in traditional societies, Cambridge: Cambridge University Press
Goody, J 1986. The Logic of Writing and the Organization of Society. Cambridge:
Cambridge University Press
Havelock, E A 1978. The Greek concept of justice from in shadow in Homer to its substance
in Plato. Cambridge, MA: Harvard University Press
Heng, M. and M. Newman, 2001. Unintended Uses ofIS/IT Strategy Planning from an
Organisational Semiomiotics Perspective, in Liu, K.; Clarke, R. J.; Andersen, P. B. and R.
K. Stamper eds. 200 I, Infonnation, Organisation and Technology: Studies in
Organisational Semiotics BostonlDordrechtILondon: Kluwer Academic Publishers,
Chapter 2, pp. 49-65
Hirschheim, R A 1989. Office automation - a social and organizational perspective. New
York: Wiley.
Innis, H.A. 1972. Empire and Communications, revised edition. Toronto: University of
Toronto Press
Kling, R 1987. Defining the boundaries of computing across complex organizations, in R J
Boland and R A Hirschheim (eds.) Critical issues in infonnation systems research, New
York: Wiley
Kling, Red. 1996. Computerization and controversy: value conflicts and social choices, 2nd
edition. London: Academic Press
Langefors, Borje, 1973, Theoretical Analysis of Infonnation Systems, Studentlitteratur,
Auerbacher
Liu, K. 2000. Semiotics in infonnation systems engineering. Cambridge: Cambridge
University Press
Liu, K, Clarke, R J, Andersen, P B and Stamper, R K 2001. Infonnation. Organisation and
technology: studies iin organisational semiotics. Dordrecht: Kluwer Academic Publishers
Luria, Aleksandr Romanovich, 1976. Cognitive Development: its Cultural and Social
Mitchell, W J 1994. When is seeing believing. Scientific American, Feb 1987
Nahapiet, J and Ghoshal, S 1998. Social capital, intellectual capital, and the organizatioal
advantage. Academy of Management Review, vo1.23, no.2, p.242-266
North, D 1981. Structure and change in economic history. New York: Boston
Putnam, R D 1993. Making democracy work: civic traditions in modem Italy. Princeton, NJ :
Princeton University Press
Sanders, P. 1982. Phenomenology: a new way of viewing organizational research, Academy
of Management Review, vol.7, no.3, p.353-360
Stamper, R 1973. Information in business and administrative systems. New York: Wiley
Stamper, R. K. 1996. Signs, Norms and Information Systems, in Holmqvist, B.; Andersen, P.
B.; Klein, H. and R. Posner eds. 1996, Signs of Work: Semiosis and Information
Processing in Organisations Foundations of Communication and Cognition (Grundlagen
Der Kommunication und Kognition) Series, Berlin: Walter de Gruyter
Stamper, R. K. 2001 Organisational Semiotics: Informatics without the computer? in Liu, K.;
Clarke, R. 1.; Andersen, P. B. and R. K. Stamper eds. 2001, Information, Organisation and
Technology: Studies in Organisational Semiotics BostonlDordrechtILondon: Kluwer
Academic Publishers, Chapter 5, pp. I 15-171
APPENDIX 1 QUESTIONNAIRE OUTLINED

I. Is the model easy to understand?
2-3. Give three examples ofIT's impact on each of the levels: semantic, and social.
4-6. Give two examples of the functioning ofIT-based systems on each level.
7 & 8. What are the strong / weak points of the model?
9. Can you use the model to examine the social effects ofIT after a time lag?
10. Can you use the model to study the unintended consequences of IT's impacts?
II & 12. It is likely that IT would interact with other technologies to produce social
impacts. How does the model handle the phenomenon?
APPENDIX 2 RESPONSE
Stamper's Organizational Semiotics Model
I. General agreement that the model is not difficult to understand. Students found the
hierarchical arrangement important and meaningful.
2. Three examples ofIT on the semantics level:
(a) Nowadays, images are easily manipulated so that a picture is more suspect as evidence.
(b) Well controlled electronic mail can displacea handwritten signature for concluding a
transaction. (c) Sometimes people consider computers more reliable than humans.For
example, for eye surgery, the fact that the laser is guided by computer may be more reassuring
than the competence of the surgeon.
3. Three examples of IT's impact on the pragmatic level: (a) Some conversational
protocols normally only face-to-face may be performed over the internet, for example,
following university courses. (b) I can communicate with virtually everyone with access to
the web. (c) Registration techniques reduce the load on memory and improve concentration
on what is at stake in the conversation.
4. Three examples of IT's impact on the social world level. (a) The Internet is driving
changes to legislation, for example on pornography and intellectual property. (b)
Digitalization speeds up everything so that, for example, an e-mail prompts an answer within
a shorter time frame than a letter. This is a clear shift in social expectations. (c) New
communities are founded online with closer relationships than between physical neighbours.
5. Two examples of system funtions on the semantic level:(a) When designing an interface, a
programmer should consider all the meanings people will associate with the signs it uses. (b)
Because information can easily be manipulated in digital form, technical strategies must
replace the redundancy of paper-based records.
6. Two examples of system functions on the pragmatics level:(a) What we understand by
"user-friendly" is largely determined by the normal rules of conversation. (b) The system
must be designed so that users can express and communicate their intentions.
7. Two examples of system functions on the social world level: (a) Systems must
technological standards that the community shares even though they may not be the best in
theory. Thus COBOL is still in use! (b) Although it is possible to store a huge amount of data
on a person, this is disallowed by laws to protect privacy.
8. Strong points: The model seems to encompass every level of human information
processing and understanding. For example, it treats the technical platform: transmission of
signals and physical media used. It deals with the understanding and intentions of the message
sender and of the interpreter.
9. Weak points: The way the layers are organized may be confusing. For example one
student read as significant the length of the strip on each level and asked why the physical
layer is so much longer than the social one, is it supposed to be more important?
10. Social effects of IT after a time lag? Yes, take the state of affairs before and after a
new IT application is introduced. This presents no extra problems and the two analyses can be
compared.
II. Reveals unintended consequences of IT? Comparing situations at different times
without a checklist like this framework, one can easily overlook areas where changes are not
obvious.
12. Phenomenon ofIT interacting with other technologies to produce social impacts: this
may be located on the IT Platform layers or, as in the interaction of film and TV with Internet,
as discussed earlier.
Chapter 9
Implementing IT in Production Settings

Organisational Sense-Making and Psychological Stress
Steven Verjans
1. SETTING THE SCENE
In 1996 a large Danish manufacturing company - Omikron25 - started a

European-Commission financed information technology (IT) research
project - BlueTech l - in order to develop and test new software that would
allow shop floor workers to have more influence and control over their daily
work in production. This new software was expected to raise the workers'
motivation and commitment and - in the long run - their productivity. The
whole idea was to use BlueTech as an experiment that should show that IT
can act as a catalyst for organisational change. Omikron had a very
traditional top-down hierarchical functional organisation, with foremen and
supervisors telling their colleagues what to do when and how. Most workers
had a career of more than 20 years with the company, some even more than
40 years, a period in which a lot had changed technologically, but not
organisationally or culturally. BlueTech, on the other hand, was
implemented by a very young research department specialised in developing
advanced robot systems and robot software, a department which was
something of an 'outsider' within a company focussed on hands-on
production.
25 For reasons of confidentiality, the names used in the paper are aliases.
180 Steven Verjans
When I entered the project in 1997, the conceptual system design was
well underway. A consultant company had developed an organisational
design for the production workshops, which was a structure with
autonomous working groups, where the different groups work in mutual
competition on their production tasks. The BlueTech software would then
support a kind of stock-exchange scenario, where each team could make a
bid on various tasks to be performed, depending on the team's workload, the
experience of its members, etc. The software would also keep a log of all
activities within each team, allowing the group to evaluate its own
functioning over time, and in comparison to other teams, and creating a
system of bonuses depending on the team's productivity. In other words, the
teams would have control over their planning, their productivity level, and -
in the long run - their own pay (cf. Verjans, Mogensen, & Lynggaard, 1998,
for a detailed description of the software).
However, this design was developed within the research department as
part of an experimental scenario, i.e. there was no intention of applying
BlueTech software in production, nor had there been any debate within the
organisation about actually changing into a structure with autonomous
groups. Therefore there was no relation with the existing situation, nor with
the envisaged situation by the time the software would be ready. There was a
clear misfit between the experimental organisation and the actual top-down
hierarchical structure. There was also a misfit between the expected self-
managing, self-evaluating workers and the actual workers' psychology, i.e.
being used to taking orders, not showing initiative, etc. So, finally, when the
project financiers demanded that the software be demonstrated in real
production settings, the conceptual design had to be redone.
2. POSITIONING THE RESEARCH
The little story in section 1 sets the scene for the research to be described
in this paper, namely "What is the relationship between new information
technology, organisations and the individuals within the organisations that
will have to use the new IT?"
Within different areas of science, a lot of effort has gone into research on
the design and development of information technology (IT) that is used or is
to be used by people in organisations. However, one cannot help feeling that
these efforts are both (a) fragmented in that only one or two factors are used
to explain empirical variation, and (b) that they do not capture the complex
whole of issues that is involved when new information technology tools
change the content of people's jobs and the functioning of organisations.
Implementing IT in Production Settings 181
The aim of my research is to develop a more holistic theory that links the
three main classes of concepts together. In contrast to much research within
the information systems literature, I assume an underlying non-
directionality, dynamism and mutual influence between IS
design/implementation, work psychology and organisational parameters.
This mutual influence is similar to structuration theory (Giddens, 1979;
Giddens, 1984), as applied to technology-related organisational research
(Barley, 1986) and information systems research (DeSanctis & Poole, 1994;
Orlikowski & Robey, 1991).
The theory described here centres around the concepts of sense/meaning,
stress and change at two levels of analysis, the individual and the
organisational. These main concepts fit into existing theory of person-
environment fit on the individual level (Caplan, 1998; Edwards, Caplan, &
Harrison, 1998) and mirrors this theory to the organisational level where it
becomes a theory of organisation-environment fit, as will be described in
section 3.
The theoretical model was developed in an iterative process, in which
intuitions and ideas that would emerge from the BlueTech design project
were related to extant literature, and vice versa. This bouncing back and
forth between literature and action provided an interesting research
technique. The next sections of the paper describe the theoretical model, the
empirical research design and instruments, and finally provide some
preliminary results from the application of a minor software tool within the
BlueTech project at Omikron.
3. THEORETICAL MODEL
The crux of the theory is that new information technology only makes
sense when it 'fits' into an individual user's conception of his world, and
when it 'fits' into an organisation's conception of its world. An individual's
interpretation of his working situation will only make full sense when all
aspects of his life fit together, i.e. when the individual's needs, capacities,
health, etc. fit with the organisation's supplies and demands, with the
individual's private situation, but also with the technologies that he uses in
his daily activities. At the organisational level a similar reasoning applies: an
organisation will interpret its situation as meaningful/sensible, when its
structure, its needs, capacities and resources fit with the environment's
supplies and demands.
The fact that the theory builds on concepts from psychology and applies
them to the organisational level of analysis coincides well with a remark by
Staw, who notes that "psychologists tend to concentrate on 'micro'
182 Steven Verjans
organizational behaviour, which includes attitudes, motivation, absenteeism

and work stress. Sociologists focus more on 'macro' phenomena like
organizational structure, strategy and environment. Staw argues that
psychological theories may help explain 'macro' phenomena, partly because
the actions of organizations are often really those of individuals or small
groups of people." (Staw, 1991), as cited in (Arnold, Robertson, & Cooper,
1995, p. 42).
3.1 Person-environment fit theory: Basis and Extensions
Person-environment fit (PE) theory (Edwards et aI., 1998) is a theory

that was developed within work psychology, more precisely within the area
of stress- and well-being related research. It "offers a framework for
assessing and predicting how characteristics of the employee and the work
environment jointly determine worker well-being" (Caplan, 1998, p.34.l5).
It is important to note that both Caplan (1998) and Edwards and colleagues
(1998) describe the theory as a framework. Indeed, the actual dimensions
with which you conceptualise or measure person and environment are not
described.
Objective Subject ive environment
Strains l linus
Subject ive person
Figure 1. A model of stress as person-environment fit. Based on Edwards, et al. (1998, p.29)
PE fit theory views humans as having a number of needs and a number of

abilities. The environment, on the other hand, is seen as containing supplies
for the individual's needs and as placing demands on the human, which can
be met by the individual's abilities. The theory states that strain arises when
the human subjectively experiences a misfit (or a 'poor fit') between needs
and supplies, or between demands and abilities. These strains serve as risk
factors for subsequent illness. The PE fit model places great focus on the fact
that strains are caused by the individual's subjective perception of fit. Firstly
because objective fit is impossible to measure and secondly because it
contributes to the explanation of why two different people in similar working
conditions may experience different levels of strain. The PE fit theory is
presented in Figure 1.
In the theoretical model described in this paper, I have enhanced this PE
fit model of stress with Schabracq and Cooper's (1998) phenomenological
concept of stress, some of Weick's (1995) ideas on sensemaking, and James
et al. 's (1990) work on meaning of organizations.
Schabracq and Cooper's (1998, p. 626-7) integrative framework builds on
two main concepts when discussing stress, namely situations and integrity.
Situations are defined as "familiar, shared cultural spaces, that enable their
actors to fulfill their recurrent needs (for example, for safety, control,
achievement and further development) in standardized ways, which are
repeatedly enacted with the help of integrative scripts. The concept of
integrity is related to what the authors call niches: "Put differently, adequate
human functioning is characterized by a specific ecology: an individual
develops a relatively stable niche to live in. Such a niche resembles animal
territory, and disturbances or infringements trigger similar stress reactions or
resistance amongst humans and other mammals, even if the interventions are
well meant. Living in such a niche allows people to build a semi-permanent
system of assumptions, meanings, images, goals, rules, procedures, skills,
etc. Schabracq and Cooper also relate integrity to signification as an active
social process. A major prerequisite for social functioning is the presence of
shared meanings. But humans will try at first to protect their integrity in
interpreting meaning of external actions and events, in a way that is similar
to the defence mechanism depicted in Figure 1. This process of signification
is a major part of my theoretical model, especially when it reaches its ideal
state: "an integrated, consistent system of meanings, which serves one's
goals in life, is open for further evolution and, at the same time, is attuned to
the prevailing meanings and needs of the relevant others" (op.cit., p.630).
Schabracq & Cooper's framework shows a striking resemblance to
Weick's (1995) concept of sensemaking Weick notes that sensemaking
mainly occurs when predicted or expected events do not happen, i.e. when
things happen that do not immediately fit into the individual's framework.
Weick also describes sensemaking as both an individual and reciprocal
(social) process: "Sensemaking is grounded in both individual and social
activity, and whether the two are even separable will be a recurrent issue in
this book, because it has been a durable tension in the human condition"
(op.cit., p.6). A similar discussion is also found in James and colleagues'
(1990) chapter on the meaning of organisations: "In summary, meaning
refers to attempts to make sense out of what is occurring in an environment;
184 Steven Verjans
it is the interpretative aspect of cognition and perception. And, by definition,

it is phenomenological" (op.cit., p. 47). James et al. relate this concept of
meaning to person-environment interactions, but they mainly use it in their
description of the psychological climate of an organisation, i.e. on a supra-
individual level. Psychological climate is conceptualised as a single, higher-
order general factor of meaning, "for judging the degree to which the
environment is personally benificial versus personally detrimental
(damaging, painful) to one's sense of well-being." (op.cit., p.53). See also the
chapter by van Heusden and Joma (this volume), who use the term
representation in a way which is similar to the term signification used here.
3.2 Theoretical model on the individual level of analysis
The resulting theoretical model, starting from PE-fit theory and extended
with the concepts described above, can be graphically represented as in
Figure 2.
Indillidual parameters
Needs
Derronds
Influence
Social relations, etc. Sens. / Meaning
Stability
Organisational Commitment
Satisfaction
parameters
Clirrote
Change
Manageme.nt style
Participation, etc.
Strain / Stre••
Instability
Technology Dissatisfaction
parameters Health
Attitude.s
Use
Participation, etc.
Figure 2. Individual-level model of sense and stress
Sense/meaning in the theory is conceptualised along the lines described

above. If an individual experiences his presence and activities in his (work)
environment as fitting to his needs, abilities, and so on, his (work) life is
deemed to make sense, to be meaningful, to constitute integrity. If, on the
other hand, the individual experiences an imbalance or misfit, then his life
may not seem to make perfect sense or be fully meaningful, and he may
experience a certain amount of strain. If this imbalance/misfit exceeds a
certain threshold, the individual may no longer experience well-being and
become stressed. In a sense well-being and stress are extremes on a scale. On
that scale, the position of the thresholds between well-being, strain and stress
will be different for each individual, but may also depend on the exact
environmental parameter that causes strain, on the general health condition
of the individual, on the synchronous presence of several imbalances, etc. In
line with the crucial role of subjectivity in the experience of strain and stress,
I want to argue that stress is not only caused by a perceived high
amount/degree of strain, but also by strain(s) that last over an extended
period of time or strain( s) that are not envisaged to end within a foreseeable
future.
PE fit theory uses a V-curve to represent the relation between person-
environment fit and strain. Figure 3 represents such a V-curve, adapted to
the theoretical model in this paper. When an individual experiences a fit
between himself and his environment, at the lowest point in the curve, his
(work) life is experienced as making sense. When there are one or more
perceived imbalances, he may experience strain. Once a certain threshold is
exceeded, this strain is experienced as stress. The ball on the curve indicates
a hypothetical individual, who is positioned above the stress threshold, due
to the environment being larger' than the person, e.g. because the
environment places too high demands on the person, or because the
environment fails to supply the needed job security. The rolling ball also acts
as a metaphor for the propositions that (a) an individual's position on the V-
curve is not fixed but mobile, and (b) that individuals - in accordance with
the integrity concept - will tend to revert to a situation of fitlbalance, i.e. the
ball will tend to roll back to the lowest point on the curve. The mechanism
that gets the ball to move, is change.
Stress
Strain
Sense
P< E P = E (Fit) P> E
Figure 3. Hypothetical U-shaped relation between sense/stress and PE fit
Change is the process that mediates between sense and stress, in so far as
meaningful situations may change and become senseless, or vice versa.
Changes of this kind can be caused by the individual's environment, by other
186 Steven Verjans
people or by the individual himself. Different types of individual change can

be distinguished: (a) individuals can try and change the environment (if they
have any influence on it), (b) they can try to cope with the misfit (for
instance by taking courses to handle new technology), or (c) they can -
through signification and defence mechanisms - try to re-interpret the misfit
in such a way that they no longer subjectively experience it as a misfit. The
specific type of change that I focus on is organisational change brought
about by new information technology.
When one limits one's focus to the development and implementation of
organisational information systems, three classes of parameters are expected
to influence the way in which individuals make sense of their activities and
environment, as depicted in Figure 2, individual parameters that influence
an individual's role in an organisation, how individuals perceive the
organisation of which they are a part, and how individuals perceive the
existing and new information technology that they use in their jobs. These
sets of parameters all need to somehow fit together into the individual's
image of a meaningful job. As mentioned above, it is hypothesised that an
individual will try to keep a meaningful configuration stable (cf. Schabracq
& Cooper, 1998's notion of integrity). Similarly, I hypothesise that
individuals who perceive their work as making sense, will experience higher
job satisfaction and organisational commitment. In contrast, individuals who
can not make sense of their job situation, will be less committed to their
organisation, show less job motivation and satisfaction and may in the long
run expenence stress, which may lead to problems with their health and
well-being.
I1Jman aspects
Climat.
Management style
Strr.ss, etc.
Organisational
parameters
Structure
Change
Environment
Size, etc.
Strain IStre ..
Instability
IT parameters Disagreement
Typos of IT
Rigidity
Participation, etc.
Figure 4. Organisational-level model of sense and stress

3.3 Theoretical model on the organisational level
On an organisational level, one may argue that a similar process of

sensemaking or meaning-creation exists, as depicted in Figure 4. This
duplication of the extended person-environment fit theory is similar to the
work by Weick (1995) described above, but it adds the concept of strain and
stress.
In order for an organisation to function, the organisation's activities need
to fit to the environment, the existing technology26, the people in it, etc. This
concept of organisation-environment (OE) fit is well established III
organisational science.
The people that make sense of the environment on behalf of an
organisation are the different groups of stakeholders. In the sensemaking
process, a major role is played by the owners (often represented by a board
of directors), senior and middle management, since they are usually the ones
to initiate actions related to the environment (decision-making). The other
stakeholders, such as trade unions, employees, suppliers, customers, minor
shareholders, or media usually play a less action-oriented role in the
sensemaking process. The three classes of parameters that are relevant in my
research concern human aspects, organisation aspects, and IT aspects.
Again, the idea is to take a holistic perspective, and look at the interplay
between these parameters, and the interpretation that the different groups of
stakeholders have on its functioning within its environment. The existence
of different interpretations within an organisation is an integral part of the
sensemaking processes within an organisation (cf. also van Heusden en
lorna in this volume). Weick (1995, p. 5) quotes Feldman (1989), who talks
about sensemaking as an interpretative process that is necessary "tor
organizational members to understand and to share understandings about
such features of the organization as what it is about, what it does well and
poorly, what the problems it faces are, and how it should resolve them." At
the organisational level, it is not enough for one group of stakeholders to
perceive a fit between an organisation and its environment. Since
organisations are collectives, it is vital that there is some agreed
understanding of the OE fit amongst all relevant groups of stakeholders. In
the model depicted in Figure 4, an organisation's environment and actions
are interpreted as meaningful if the stakeholders have a minimal shared
understanding of the organisation's current status and direction.
Organisational sense/meaning can thus be regarded as the common
26 Technology is to be interpreted as the whole range of activities, processes and machines

that are used by an organisation to create its products or services, and perform its
activities.
188 Steven Verjans
denominator of all stakeholders' interpretations of OE fit. Again, stability is

related to sense, as it was at the individual level. In accordance with
Schabracq and Cooper's (1998) integrity framework, one could argue that
organisations, like individuals, prefer to operate within their familiar, well-
know niche.
The distinction between objective and subjective OE fit is not as clear-cut
as on the individual level. One could well imagine a situation where
organisational members and owners perceive a subjective OE fit, but where
'outsiders' clearly perceive a misfit, but this kind of situation will be very
rare. On the organisational level, a subjective perception of fit is in itself an
inter-subjective perception, because there is more than one 'perceiver'.
Figure 4 also introduces two new concepts, namely organisational strain
and organisational stress, in analogy with individual strain and stress.
Organisational strain arises if the organisation and its environment are
experienced as being unbalanced, for example when there is a misfit
between the complexity of an organisation's technology and the training
level of its employees. In analogy with individuals, organisations can be
conceived of as having needs and abilities that need to be fitted to the
supplies and demands from the environment. On the other hand,
organisational strain can also arise if there is disagreement between the
different stakeholders to such a degree that no common vision can be
reached.
Organisation stress is then defined as caused by one or more
organisational strains that last over an extended period of time, are present in
a 'large enough' degree to exceed a certain threshold, or are not envisaged to
end within a foreseeable future. As on the individual level, one can imagine
a V-shaped curve representing the relation between organisational strain and
OE fit, with different organisations having different thresholds. In analogy to
individuals, it is hypothesised that organisations may experience health
problems because of persistent stress conditions. Health problems at an
organisational level may be reflected in declining turnover or profits,
downsizing, staff turnover, etc. If such a situation occurs, organisational
members can perceive of the organisation as not making any sense at all.
In the literature, there has recently been quite some attention to
sensemaking regarding information technology in organisations (Fulk, 1993;
Orlikowski & Gash, 1994; Prasad, 1993; Weick, 1990). That research goes
very much along the lines as suggested in my theoretical model. I perceive
the difference between these contributions and the current as one of focus
and generality. The existing contributions focus mainly on the group and
organisational levels of analysis, whereas I argue that the individuallevel of
analysis also plays a major role in acceptance and use of technology.
3.4 Practical application of theory
In this section, I want to briefly sketch how the theory described in the
previous sections can be practically applied during a project of IT design,
development and implementation.
The situation from which the analysis starts, is one where someone
within an organisation has decided to develop and implement a new
computer-based information system (IS) that is to have certain
characteristics. A change agent is then appointed who is to co-ordinate the
design, development and implementation of the new technology. Figure 5
shows a temporal overview of the different stages at which my theory can
inform the change agent.
¥
Users' psy- Exisfing Character-is f-P~~t;,I09;:' IS design &
chologicol infol"rmtion -fics of new " col change: devment Psychologi- IS result
[ col result
pOr'OrtEtus sysw:ms IS
L."~.,~}-._Cfi_'~_fi'_5___
y----.<~0
r-------l
Orgonisa- [ Qoganisatio-I
Organisatio-
tional : 1'101 change 1
nal result
parom£ters
L~:ti!!~~_l
Misfit onolysis Misfi t analysis Misfi t anolysis Misfitonolysis
Ini1ials1ote Envisaged sta. Changll!:stoie Result sm1'e
Figure 5. Graphical representation of the analysis model. The circles represent different states
in a design project. The broken lines indicate activities that are not inherent in every project
When the change agent enters the picture, there is a certain situation in
the organisation that he can analyse for possible misfits, by looking at (a)
psychological parameters of the group of users and other stakeholders, (b) a
number of organisational parameters, and (c) existing information systems
(computer-based and others). Based on the results of the initial misfit
analysis the change agent can then analyse the intended characteristics of the
new computer-based information system, and see if 'the system'might solve
existing misfits or problems and/or introduce new misfits or problems. One
could label this state the misfit analysis of the envisaged state, i.e. how one
imagines the organisation, the psychological state and the information
system to be after the system' has been introduced. This is presented as a
new state - s' - because it represents a changed perception on account of the
change agent. At this stage, the results of the analysis are probably best
discussed with relevant parties, because important decisions need to be made
at this point in time. If the agent has discovered potential new misfits based
on his analysis, he needs to make sure that all parties are aware of those, and
are intentionally pursuing that course. On the other hand, the parties may not
190 Steven Verjans
be aware of new misfits, and may decide to make changes based on the
envisaged misfit analysis, either at the organisational level or at the level of
the intended characteristics of 'the system'. Finally, the parties may decide to
drop the new development project altogether.
State s" assumes that one has decided to pursue the development of a new
system, and has started a number of activities to design, develop and
implement the new IS. As the broken lines indicate, the IS development
activities mayor may not be accompanied by organisational or
psychological change activities. Depending on the duration of the
development and implementation project, one may choose to perform a
misfit analysis 'underway', i.e. while a number of change and development
activities are ongoing.
Finally, the change agent can perform an analysis at the result state - s"'-
which is some time after the formal change and development activities have
finished. In organisational settings this analysis is usually rather informal in
order to establish whether or not the project was a success, with the success
criterion being: Does the system work? Using the theoretical framework in
this paper, the success criteria can be psychological, organisational as well as
system-related. The causal links between the different classes of parameters
and the success of a new IT system are graphically represented in Figure 6.
Psychological parameters
ISD Approach
r----------------~
Organisational parameters
Success of new IS
Existing IS parameters
--~ ____ ~~~~~~~~ ____r--~
I
Characteristics of new IS - - - - - - ' - - - - - - - - - - - - - - - - - - - '
Figure 6. Causal links between the different classes of parameters in the theoretical model.
Again the broken line indicates optional activities
The psychological, organisational and IS parameters are expected to have

both a direct and indirect influence on the success of a new Information
System. All the parameter classes on the left-hand side of the drawing are
hypothesised to influence the IS development and implementation (IS D)
approach. Those parameters will also determine whether a complementary
organisational development (OD) approach is needed for a particular type of
IS, and they will then have an influence on that. Those left-hand parameters
that are not changed by the development and implementation are
hypothesised to have a direct influence on the success of the new IS. For
instance, an organisational climate with a high tension that is not changed
during the ISD project is expected to determine the acceptance and use of a
new system. On the other hand, the ISD and OD approaches are also
expected to determine the success of the new system.
But what determines success in this theoretical model? Firstly, an ISD
project is deemed technically successful if the information technology has
the required functionality and meets a number of standards regarding
effectiveness, efficiency, user-friendliness, etc. One could argue that
technical success also depends on whether the software developers feel
satisfied that they have done a good job. Secondly, an ISD project needs to
be organisationally successful. The least that an organisation aspires when
introducing a new information system, is that the IS does not create a misfit
with the organisation's perception of sense/meaning. More often, though, an
organisation expects an information system to solve some existing
organisational strain or stress, i.e. an information system is deemed
successful if the general organisational strain level goes down. Finally,
similar success criteria apply at the individual level of the end-user. The least
that organisational members expect is that a new information system will not
introduce a misfit into their (work) life. Again, one may hope that an
information system will do more than just avoid creating misfits. Indeed,
information systems are usually developed in order to support the work of
organisational members in such a way that they lower the level of strain, e.g.
by streamlining certain boring or superfluous tasks.
Ideally the success of an IS project hinges on the three classes of criteria.
In practice, however, one will very often find that the organisational and
individual success criteria do not coincide, or that they are even orthogonal.
A major claim of my theoretical model is that situations where there is a
large degree of individual strain in an organisation will automatically lead to
a raised level of organisational strain. Therefore, the maximum success of an
ISD project will be achieved when the project is a technical, individual and
organisational success.
An important final remark is that any ISD project that involves changing
the organisational or individual routines or procedures will at first be
experienced as a threat to organisational or individual integrity, and thus
cause a raised level of strain. This raised strain level is not necessarily
negative, in so far that it can trigger a signification or learning process.
However, if the strain level does not go down after some time (usually it
takes a few months to get used to completely new technology), one can start
to have doubts about the success of an ISD project.
192 Steven Verjans
4. EMPIRICAL RESEARCH
The theory described in section 3 above is extensive and includes a lot of

entities and variables, so it is quite a challenge to devise a research design
that allows one to gather empirical data to test the theoretical claims.
Moreover, some of the variables and concepts are quite hard to
operationalise, e.g. concepts like sense and meaning. The literature on
sensemaking does not contribute very much in this respect, since much of
the literature is purely theoretical (e.g. Schabracq & Cooper, 1998; Weick,
1995).
An interesting exception is the chapter by James, James and Ashe (1990),
who distinguished and empirically tested 17 perceptual measures of the
meaning of work that clustered together in 4 factors informed by work from
Cooper, Sloan and Williams (1987) and Hackman and Oldman (1975, 1980).
In developing research instruments, I have used the same technique,
namely borrowing measures and concepts from other existing instruments.
The next paragraphs describe (a) the methodology used in gathering data and
(b) the creation of research instruments.
4.1 Methodological considerations
A first remark about the research method used here is that the main aim
of the project was not to test elaborated and well-defined hypotheses, but to
explore the relations between research variables. Moreover, because I
perceive information systems development projects as very complex, my
aim is not to distinguish one or two factors that can predict IS development
success or failure. As described in §3.4 above, the research clearly takes a
process orientation. More specifically, I characterise my research as being a
process-oriented exploratory theory (see Franz and Robey 1987).
When it comes to choosing a method for doing process-oriented
exploratory research, it is clear that the survey method is not well-suited,
since I want to study phenomena in depth in their organisational context, and
not through a standardised technique across a representative sample of
organisations. The (field) experiment method is not appropriate either,
because it is too intrusive and artificial to be used in evaluating existing
sensemaking processes. The method I have chosen is a multiple-case, in-
depth and longitudinal approach, although it also contains a flavour of action
research. I described elsewhere how the ideas that gave rise to the theoretical
model above were generated through an interactive process involving both
my active participation in the BlueTech project and concepts and
frameworks in extant literature. I want to argue that this type of participatory
action research (Whyte, 1991) was a first phase of 'idea generation'. In the
meantime, my active participation in the BlueTech project had ended. It

forms a pilot case study for the research, in that it is mainly used to generate
ideas, test techniques and research instruments and get preliminary feedback
on some of the concepts used. Moreover, due to practical reasons it is not
possible to perform a longitudinal analysis at Omikron. My second case at
Alpha Industries is then used to substantiate some of the results from the
BlueTech case, and to apply them in a longitudinal setting.
This research design uses Kathleen Eisenhardt's classical (1989)
'roadmap for building theory from case study research' as a guide. That
roadmap synthesises previous work on qualitative methods (e.g., Miles &
Huberman, 1984), the design of case study research (e.g., Yin, 1984), and
grounded theory building (e.g., Glaser & Strauss, 1967). The roadmap,
presented in Table 1, distinguishes seven main steps in the development of a
research strategy for exploratory case studies.
Table 1. Process building theory from case study research (adapted from Eisenhardt, 1989)
Step Activity
1. Getting started Definition of research questions

Possibly a priori constructs
Neither theory nor hypotheses
2. Selecting cases Specified population
Theoretical sampling
3. Crafting instruments and protocols Multiple data collection methods
Qualitative and quantitative data combined
Multiple investigators
4. Entering the field Overall data collection and analysis
Flexible and opportunistic data collection methods
5. Analyzing data Within-case analysis
Cross-~ase pattern using divergent techniques
6. Shaping hypotheses Replication, not sampling, logic across cases

Search evidence of "why" behind relationships
7. Enfolding literature Comparison with conflicting literature
Comparison with similar literature
8. Reaching closure Theoretical saturation when possible
In my research, the result of step 1 is the theoretical model described

above, which builds a set of 'a priori constructs' with some of the relations
between them. It is not a full-fledged theory in the scientific meaning. In the
second step - selecting cases - the population was specified as Danish
production companies within the metal sector'. The selection - theoretical
sampling - of these two companies was based on (a) practical considerations,
and (b) the fact that they differ in size, but also that the IT systems are
different. The third step of the process resulted in the data collection
194 Steven Verjans
techniques described in the next paragraph, which combines qualitative and

quantitative techniques. Where possible, I also asked colleagues to
participate in the data collection phase (mainly the interviews).
4.2 Organisational-level research instruments
As described above, the theoretical model is quite extensive, and required

quite some effort to operationalise. At the organisational level of analysis,
the main instrument is the OrgCon®, an expert-system tool based on Burton
& Obel's (1998) framework for strategic organisational diagnosis and design.
The OrgCon® - Organizational Consultant - builds on a multi-contingency
theory concept, informed by an information processing view on the firm.
Based on an analysis of the answers to 60 questions, the OrgCon® can
determine whether the design of an organisation fits with its environment,
technology, strategy, climate etc, or whether there are one or more misfits.
Recent longitudinal research by Burton and colleagues (Burton, Lauridsen,
& Obel, 2000) shows that their framework can indeed predict organisational
health as conceptualised above. A survey performed among small and
medium Danish production companies shows that organisations without
misfit have a significantly better financial performance over a two-year
period, than those with one or more misfits. Using OrgCon®, one can also
determine whether different (groups of) stakeholders hold different views on
an organisation'S well-being. In another project, Dejbak and colleagues
(Dejbak, Nissen, & Obel, 2000) used the analysis of OrgCon®-based
interviews of all top managers within 5 Danish companies as a basis for
discussion within the management team, in order to 'align' the managers'
views with each other.
The answers to the 60 OrgCon® questions are gathered during structured
interviews. Next to the OrgCon® analysis instrument, I have developed
instruments for specific organisational topics:
- A small questionnaire was developed for managers, software developers
and trade union members, in which they are asked to specify what they
think employees will answer to some of the individual questions
regarding climate, satisfaction, commitment and meaning. This allows
me to compare average values of employee answers to the perceptions of
other stakeholders.
- A small interview guideline was developed for asking IT managers and
software developers about their views concerning existing information
technology systems within organisations.
4.3 Individual-level research instruments
In contrast to the organisational level - where I could use an existing

instrument - a lot of work went into the development and testing of a
questionnaire instrument at the individual level. The instrument contains
selected entries from a number of existing validated questionnaires, but also
some new items. The final instrument is a 19-page document and contains
some 200 items. It went through numerous revisions and was validated by
Omikron employees before its last revision. Table 2 shows an overview of
the items in the questionnaire together with their source.
Table 2. Overview of the items ofthe individual-level questionnaire
Individual parameters # itcrr Source

Needs in ideal job 18 VSM
Supplies in current job 18 sv
Demands 13 AMI
Demands fit/misfit 13 sv
Influence 8 AMI
Information & goals AMI
Meaning/variationllearning AMI
Anxiety / insecurity 4 AMI
Colleagues & management 7 AMI
Commitment 5+9 sv
Job satisfaction 8 AMI/s'
Timing of work 4 AMI
Work-home interface 2 AMI
Subjective health 12 AMI
Stress indicators 12 AMI
Organisational parameters
Climate 7 OrgCon
Management Style OrgCon
Organisational strain I sv
Organisational parameters 12 EOS
Information systems
Use of technology 4 sv
Technological skills 9 sv
Attitude towards IT 8 sv
Use of new IS 2 sv
Evaluation of new IS 10 kbr
Participation during desvelopment ofnew IS kbr
Demographic items 6 AMI
In Table 2, VSM is the Danish version of Hofstede's 1982 Values Survey

Module based on (Hofstede, 1980). AMI is an instrument developed by the
Danish National Institute of Occupational Health (Arbejdsmilj0institut) for
measuring psychological climate (Kristensen, 1999). This instrument is
196 Steven Verjans
mainly based on Karasek's Demand/Control model of occupational stress.

OrgCon is the instrument developed by (Burton & Obel, 1998) as described
above. EOS is the Employee Opinion Survey (1992) developed at the
Institute of Work Psychology at the University of Sheffield. All items with
other codes were developed by myself or my colleague at the University of
Southern Denmark.
Next to this major questionnaire, interview guidelines were developed
measuring (a) some of the novel concepts in the theoretical model, such as
sense, meaning or organisational stress, and (b) major organisational and
technological changes in the recent past.
4.4 Data collection and respondents.
The data collection phase at Omikron concerned the Pipe Viewer module
of the BlueTech software, which is the only module of the ambitious
software design that was implemented. At the time of the data collection the
tool had been used in actual production during a 7-month period in the
Omikron department that produces different types of one-of-a-kind steel
pipes. This pipe workshop produces some 40 to 50,000 pipes per year. The
PipeViewer software module, which was installed on three computers on the
shop floor, allows workers to view 3D drawings of the pipes that they are
producing. Figure 7 shows a screen image of the software tool.
-
l-tt:!I=.d=..L:. ,11 ... -
",Q a ll III 0 ... ""
A.
e.
Figure 7. Screen of the PipeViewer module
The Pipe Viewer is meant to be used by the pipe smiths in the workshop,
since they are the ones that need to put the different pieces of the pipes
together. The result of their work is then fully welded by one of their
colleagues at a different location. Significantly, the pipe smiths are not
obliged to use this tool. It is regarded as an extra tool that can be used when
pipe smiths are in doubt about complex pipes, with the aim of (further)
reducing the number of mistakes in production.
As discussed above, one of the main objectives of this case study was to
gather data from different groups of respondents. Table 3 shows the groups
of respondents that participated, together with the type of data that was
collected. In total, 58 questionnaires were returned and 25 interviews were
held.
Table 3. Overview of collected data material at Omikron
Interview i
Question- Employee Organisa- I
n naire Individual perception tional IT
Top managers 4 2 4 4 4 I
Middle managers 5 4 5 5 5 2
Local managers I I I
Union representatives 6 6 3 3 3 I
I
Pipe workshop workers 42 42 II
Software develop ers 4 4 I I I I i
62 58 25 14 13 4
As suggested by Eisenhardt (1989), the research contains both qualitative

and quantitative data. Whereas all respondents were asked to complete the
questionnaire, only a relevant sample was interviewed. Only four
respondents did not complete the questionnaire. Completing the
questionnaire took between 30 and 70 minutes, while interview duration
varied between 30 minutes (pipe shop workers) and 1,5 to 2 hours (other
respondents). Part of each interview was semi-structured, following the
interview guidelines described above, while each interview was concluded
with an unstructured part.
5. PRELIMINARY EMPIRICAL RESULTS AT THE

ORGANISATIONAL LEVEL
On the organisational level, the qualitative data shows that the history of
Omikron has been characterised by crises that often resulted in personnel
reductions. This is reflected by a high level of insecurity amongst
198 Steven Verjans
employees27 , but this insecurity seems to be part of the particularly

fluctuating market that Omikron is in. Currently there is strong pressure
from especially South Korea, which is producing products below market
price. This situation forces Omikron to reduce its costs by raising
productivity, cutting on materials costs through better acquisition policies,
better planning, etc. As part of the aim of raising productivity, management
started a project in 1998 that aims to change worker participation in decision
making. The project aims to develop groups of workers that can take over
some of the responsibility for their own production activities. However,
these organisational change efforts are hindered by the traditional culture at
Omikron, where the trade unions (more than 15 different unions, one for
each profession) are very powerful stakeholders, who see the need for
changes very well and participate in the decision making, but who are also
restricted by their own historically rigid structures. Moreover, due to its large
size (approx. 2700 employees) Omikron's traditional functional
organisational structure is also rather hard to change.
5.1 Misfits within stakeholder groups
The interviews with different managers revealed an interesting

dissonance within management regarding organisational climate,
management style and other organisational parameters. Based on the
interviews I deduced the existence of two 'clans', which I labelled the
operationalist and the strategist clan. The latter believes that operational
decision making should happen at the lowest possible level, and that top
management should focus on long-term strategy and the formulation of a
clear vision for the future. The former is more inclined to centralise a great
deal of operational decision making at the top level of the organisation, and
try to streamline operations by ever better planning procedures and strict
operational control. Interestingly, the balance of power' between the two
clans changed with the installation of a new CEO in 1998, when the overall
atmosphere changed from visionary/strategic to 'very operational '. This
dissonance is supported by the quantitative data, as presented in Figure 8.
The figure shows the mean values of significant items of a cluster analysis
that was based on the answers to the organisational questions in the
interview and the questionnaire. Clusterl contains 4 middle managers and 3
27 This is supported by quantitative data, i.e. when comparing the answers to questions on
insecurity. Pipe shop workers have a score of 98.6% on the insecurity index, where the
Danish average is 50%. The complete comparative analysis for the pipe shop workers is
provided in Verjans, S. (Forthcoming)
trade unionists, whereas Cluster2 consists of 4 top managers, 1 middle

manager and 3 trade union representatives.
The members of Clusterl clearly perceive a higher degree of
scapegoating and a lower morale within the organisation. They interpret top
management as having a reactive and control-oriented management style
with a relatively low degree of employee participation in decision making.
Cluster2 members on the other hand perceive Omikron as much more
formalised and centralised than Cluster 1 members: they indicate many more
written job descriptions, much more control of decisions, and a higher
degree of management involvement in micro-level operational decisions.
These differences within management have recently led to a number of
resignations of high-tenure middle managers. Moreover, they are reflected in
management's dualistic attitude towards information technology. Whereas
one group is in favour of developing information systems to support de-
centralised decision making, the other wants to integrate as many existing
systems as possible into a centralised system where all information and
decision making can be centralised.
At least as striking is the fact that this dissonance also exists amongst the
trade unionists in my sample.
S<X+J89Xfing low HlgJ
Morde' Low HIg,
Mioo-invoIv9TB1t Low Hig,
MOl style Control MotlvdlO'l
MOl style Re:dive Prca:::tlve
Control ciEds lons ~ H::IJX,

Writtaljoodeso. ~ l~
O'g unct:r D'ESS ure Low 4.5 HIg,
Po1idp:tloo' Low HigJ
Ousta 1 (n=7) Ousta 2 (n:8)
Figure 8. Strong differences within Omikron management and trade unions, as revealed by a
cluster analysis of their organisational level answers. All differences are statistically
significant to the .05 degree, except for those items marked by an asterisk which are
significant to the .01 degree.
5.2 Differences between stakeholder groups
When looking at all stakeholder groups (top management, middle

management, unions, and workers), there are other significant differences
(see Verjans, Forthcoming). Management thinks it has a high degree of
200 Steven Verjans
credibility, while workers and trade unions think it's only moderate. Trade
unions feel that management is willing to take risks, and that Omikron is
rather formalised, whereas middle managers perceive the organisation as not
so formalised or risk taking. But there is also a disagreement between trade
union representatives and their constituents: Whereas trade unions perceive
communication at Omikron as rather problematic, workers do not do so.
The OrgCon® expert system ran its multi-contingency fit analyses on the
data collected and came up with the following situational and contingency
misfits. Situational misfits indicate imbalances between rather steady
variables in the organisation and its environment, such as environment,
technology, strategy, management style and organisational climate. The
contingency misfits on the other hand indicate imbalances between the
situational variables and the design of the organisation, i.e. its configuration,
complexity, centralisation and formalisation.
Table 4 shows that OrgCon®'s analysis of management's perception of
Omikron is more problematic than that of the trade unionist's perception,
especially if one looks at the number of situational misfits. Another striking
aspect is the high number of climate- and strategy-related situational misfits.
Within the contingency misfits, there is consistency in the complexity misfit
amongst all respondents, but the formalisation and configuration misfits
show a rather more fragmented situation.
Table 4. Overview of situational and contingency misfits determined by an OrgCon® analysis

of different stakeholders' answers to 60 questions about the organisation and its environment
Top Middle Top Middle

Manage- Manage- Trade Manage- Manage- Trade
ment mcnl Union mcnt ment Union
Situational misfits (n~4) (n~5) (n~2) Contingency misfits (n~4) (n~5) (n~2)
Org. Climate <-> Strategy J No contingency mis fits

Org. Climate <-> Management Style Configuration should be matrix
Org. Climate <-> Environment Configuration should be divisional
Org. Climate <-> Technology Complexity too low
Environment <-> Strategy Complexity too high
Environment <-> Technology Centralisation too low
Strategy <.> Market I Product spread Centralisation too high
Technology <-> Employee education Fonnalisation too low
Technology <-> Product innovation Fonnalisation too high
Average # ofrrusfits per respondent 4.75 4.6 2.5 Average #- ofnusfits 1.6 1.5
The most striking point is the large number of misfits that OrgCon®
discovered. However, Burton, Lauridsen and Obel have in recent research
(Forthcoming) noted that the number of misfits may not playa decisive role
with regard to an organisation's health. It's rather the presence of at least one
misfit that diminishes organisational performance (as measured by increase
in return on investment). The authors' empirical data also suggests that it is
the elimination of all misfits in a coherent fashion that will produce a
remarkable increase in performance, not a piecemeal or sequential
elimination of single misfits. Indeed, the elimination of one misfit may result
in the creation of one or more new misfits.
5.3 Organisational strains
Overall, both the qualitative and quantitative data at the organisational

level seem to indicate a substantive amount of organisational strain within
Omikron, internally within stakeholder groups, but also between
management and other stakeholder groups. Whether these strains have
actually risen to such a level that they are to be regarded as stress, is hard to
determine at this stage. In order to establish stress, a feedback round at
Omikron is needed where the current results are presented to and discussed
with the different stakeholder groups.
In relation to information systems development, I noted earlier that these
two opposing groups also have different views of the role of IS, which does
complicate matters for IS developers.
6. PRELIMINARY RESULTS AT THE INDIVIDUAL

LEVEL
Interviews with the local managers and the workers showed that the
PipeViewer was frequently used immediately after its introduction, when it
was still quite new. In those 2-3 months the number of mistakes also went
down. However, after 7 months the system was not being used so often and
then mainly by the more inexperienced workers. Indeed, the tool is mainly
used in cases where pipe smiths are in doubt, and the experienced smiths are
not easily in doubt, since they have been producing similar pipes for 20-30
years on the basis of a simple diagram and a list of measures. Therefore, one
of the workers commented that the number of mistakes had gone up again to
the same level as before the introduction of the tool. Most interviewees from
the pipe shop did indicate that they were happy that the tool was there, as a
safety net, just in case. However, the interviewees did not perceive a rise in
their satisfaction, motivation or commitment, nor did they feel that there was
extra stress due to the new technology. They did not feel that their work had
become more varied or more meaningful. Finally, many interviewees
regarded the Pipe Viewer software as a first start, "just to get used to the
technology" .
202 Steven Verjans
6.1 PIE fit and sense / meaning
In the interviews respondents were asked what made their work

meaningful. The answers were quite diverse and included money, the fact of
getting more influence, technological innovation, etc. There were striking
differences between managers and trade unionists on the one hand, and
workers on the other hand. The former group found meaning in the
challenges, variation and responsibilities of their work. The answers of the
latter group showed more diversity, but in general the meaningfulness of
their job lies in the quality of their work and the fact that they contribute to
the building of a complex product, that they are part of a meaningful whole.
One worker answered that the meaning of his job was "at det hele hrenger
sammen", i.e. that everything fits together: his tasks, colleagues, wages,
working hours, family, etc.
Table 5. Statistically significant (p < .05) correlations between major individual-level

variables
/'l//// ,
A:~ /, /{:.,$y-
~V~~ ~/ /
<7.~ . ;,..." .~o ,l'
*' ·.
;s-/
i,.''Y;..(,~ . ~\ ~~,?~
~\ ~~'>~~:.r"~~' .J..il~~~ ,,<>~\. "'~. .
..i§. /
~ \~ .~ 0 ~'\'. ~~ ;,
~ ~7<::f~ ,~<:$ ~f ~.fO,,09\,'f¥ ,>:§>'i:"~<I"

.. ...,."
I
~a:
Misfit : Needs > sup plies ~
I I " ,,," " " "
-c
Misfit : Demands > abilities ~
I ""
g 1Il Influence
" ~
" " " ~ ~
E 1i
..... .g
~ ~
Predictability I info
Mcanin~f"lness "" ~
~
" r;;-J "
" " E:1
" "
" " " ~ ~ ~
.9!
OrQ.nisational corrrnitmcnt
" ~
" " " I " II"1 ~ ~
Job satisfaction
'" Subjective hea lth
....... ..c~ "" ~ ~
~ " ""
" L JIi'
~ ~
"""
!l :: D Insecurity ~ ~
~g.g
fJ)J:..g " sick days ~ ~ ~ ~ I 1
Stress ~ ~ ~ ~ C
The suggestion that sense / meaning is a complex but also an integrating

concept is supported by the empirical data, as can be seen in Table 5. In
accordance with the expectations in my theoretical description above, needs /
supplies misfit is negatively correlated with perceived meaningfulness, but
also with job satisfaction and organisational commitment2 8 • The role of the
28 There has been quite some work that views organisational commitment from an exchange
perspective, i.e. as a fitting mechanism between work values and work rewards, a theory
quite similar to the needs / supplies misfit part of the theory above (e.g. Mottaz, 1988, p.
470).
other half of the PIE fit theory, the fit between a person's abilities and the
environment's demands, does not seem to get a similarly strong empirical
support. Its correlations are much weaker and fewer than those of needs /
supplies misfit, but there are some striking correlations with some of the
demographic variables such as age and tenure. This seems to imply that
ageing does not cause misfits between needs and supplies, but that it may
well create misfits between the demands that the work environment makes
and the abilities that a person has. Closely related to the concept of
sense/meaning is the meaningfulness variable 29 , which is the variable that
correlates with the highest number of variables. It is very strongly related
with needs/supplies misfits, influence, organisational commitment and job
satisfaction, but also with subjective health, insecurity and stress. The table
also shows that neither needs/supplies misfit nor abilities/demands misfits is
strongly correlated with stress, but that needs/supplies misfit are strongly
related to meaning and satisfaction, and that the latter are in tum negatively
correlated with stress symptoms and the number of sick days. Thus one may
tentatively conclude that the correlations in Table 5 lend preliminary support
to the 'sense / meaning' part of the theoretical model. When the
environment's supplies fit with (or are larger than) a person's needs, that
person is inclined to show higher job satisfaction and commitment.
6.2 Strain I stress and health
When studying the Spearman correlations in Table 5, it is interesting to

notice that stress shows no significant (p < .05) correlation with either part of
the PIE misfit, as I mentioned in the previous paragraph. Figure 9 shows the
scatter diagram of strains/stress with the two PIE fit variables, needs /
supplies fit on the left and abilities / demands fit on the right.
29 The meaningfulness variable is a measure with 7 items asking about (1) learning new
things, (2) meaningfulness of tasks, (3) important effort, (4) variation, (5) using one's
abilities, (6) extending abilities, and (7) motivation.
204 Steven Verjans
5 :
S S ":
.
T T
R R
"
I
4 •••••
00 • • •
: .:
I
N
N
S
....
:0. 0. 0.
HQflVQ
Figure 9. Scatter diagrams of the Strains/stress and PIE fit variables. The scale for STRAINS
ranges from I to 5 where I indicates high strains. The needs / supplies misfit scale (MQA VG)
ranges from -4 to +4. Negative values on the MQA VG axis indicate supplies> needs. The
demands / abilities misfit scale (RDQAVG) ranges from -2 to +2, where negative values
indicate abilities> demands.
The two scatter diagrams clearly indicate that there is no linear relation
between variables, but it also shows no support for the hypothesised U-
shaped relation between stress and PIE fit, which in these diagrams would
have been presented as an inverted U-shape, due to the inverted STRAINS
scale.
In contrast, stress does show strong negative correlations with
meaningfulness, job satisfaction and organisational commitment. However,
when performing a linear regression analysis (cf. Verjans, Forthcoming) to
find out which variables were significant determinants of stress in the small
Omikron sample, it was interesting to notice that the strongest antecedent of
stress, predicting more than 40% of its variance, was the variable Negative
feelings', a variable that measures one's psychological disposition in the last
4 weeks 30 • In that analysis, satisfaction was a weak (negative) predictor of
stress, while needs I supplies misfit had very low significance.
Another interesting fact from the Omikron sample is that stress is
negatively correlated with subjective health31 , but shows no significant
correlation to the more objective measure of health, the amount of sick leave
a person has had in the last 12 months. In contrast, meaningfulness,
commitment and satisfaction do show a significant negative correlation with
objective health. Not withstanding these correlations, the linear regression
analysis of objective health indicates that the variables in my study only
have very limited predictive capability (about 20 % of the variance) as
regards sickness.
30 The questions this variable consists of, are: "During the last 4 weeks, how often have you
felt:" nervous, down, sad, worn-out, tired, stressed?
31 Subjective health is measured by one question: "How would you describe your health in
general?" with a 5-point scale from Excellent' to Bad'.
6.3 Individual variables, organisational and

technological variables
Table 6. Statistically significant (p < .05) correlations between psychological variables and
organisational and technology variables
Js '" Org. climate: tension

"" ""
~ ~
"~
·s "jg :EOrg. climate: resistance to change ~ ~ ~ ~
"" "" " "" "

Cd 0 .I.'I:S
e!'.= !a Motivational management style
o >~~~~======~~--~~-------1------~
~
Degree of participation ~ ~
6 :tl Use of computers at job

"" "
~
-5= >':C
~.~ PC at home? ~
"
~ - ~ Negative technology attitude
Table 6 shows the correlations between individual variables, i.e. those

variables that measure respondents' psychological status, and organisational
variables that measure the perception that individuals have of their
organisation. Note that - in order to enhance clarity - this table does not
contain a full correlation matrix with all variables on both axes.
There are strong correlations between the individual- and the
organisational-level variables, which indicates that psychological well-being
does play a role in the wayan individual perceives his/her organisation.
Again, the correlations are stronger for the variables that measure sense /
meaning (the four columns to the left) than for the stress / health variables.
Interestingly, stress is not correlated with organisational climate, but only
with management style and degree of participation.
Table 6 also shows that psychological well-being is correlated with
technology variables, esp. with an individual's technology attitude.
Interestingly, the insecurity variable is related to all three technology-related
variables, indicating that those individuals that use computers and have a
positive attitude towards technology are less insecure with regard to their
future.
The fact that an individual's perception of organisational variables and
his/her technology attitude are related to meaningfulness, commitment and
satisfaction provides some support for the integrity aspect of the theoretical
model. An individual's perception of his/her situation within a work
environment hinges on a global fit involving a subjective evaluation of
206 Steven Verjans
individual needs, social aspects, meaningfulness, organisational climate,

management style as well as attitudes towards technology.
6.4 Individual differences
An important remark at the beginning of this chapter was that existing

theories hardly consider individual differences in their treatment of the
individual in his/her relation to the organisation and to technology. In the
interviews with pipe shop workers at Omikron it became clear that people
are rather different in the goals they have at work, in the way they give
meaning to their working life, the way they perceive the same organisation,
and so on, and that these differences are only partly linked to a respondent's
profession. In order to find support for this observation, I performed a
number of cluster analyses of the data, based solely on the 18 questions
measuring the strengths of respondents' needs. The analyses identified four
clusters with significant differences in psychological, organisational and
technology-related variables. These different types of people experience
different needs / supplies misfit, meaningfulness, insecurity and stress, they
have a different perception of the degree of organisational formalisation, and
have different attitudes towards technology. The strong variances between
the clusters seem to support the intuition that individual differences playa
major role in the way that people perceive and make sense of their
environment. Interestingly, the clusters are not strongly related to job
satisfaction or organisational commitment.
Another striking point about the clusters is that they are strongly related
to the different types of stakeholders, as presented in Table 7. Cluster3, for
example contains no workers, but has the majority of managers, trade
unionists and IT -developers. The workers, on the other hand are spread
across clusters 1,2, and 4. This distribution seems to support the intuition
that people with different needs will end up in different types of jobs, but
also that within one type of job, there are still rather significant differences
between individuals.
Table 7. Frequency table showing the distribution of stakeholder types across the 4 needs-
based clusters
Cluster
Jobtype I 2 3 4 Total
Top manager 2 2
Middle manager I 3 4
Trade unionist 2 4 6
Worker 21 9 12 42
IT -developer I 3 4
Total 22 12 12 12 58
7. CONCLUSIONS AND LIMITATIONS
This research started out as an exploratory attempt to define,

operationalise and find support for a holistic model of the complex relation
between individual, organisational and technological aspects of the
development and introduction of new technology. The theoretical model that
was developed combines existing research from a number of fields, and
builds on them. It is proposed that individuals have a need for an integrated
view of their functioning within their environment, and that this integrated
view consists of an overarching, relatively stable cognitive framework that
gives meaning to their functioning, and allows them to make sense of their
environment and their role in it. This process has received different labels in
the literature -sense-making, representation, signification- but they all
involve some type of fitting process, whereby individuals interpret
information and events in terms of their existing cognitive frames, and try to
fit new information into that existing framework, preferably without altering
their existing frames too drastically. It was proposed that the presence of a
balanced cognitive framework will lead to perceptions of meaningfulness
and satisfaction, and that the absence of this balance - or the existence of
misfit - will lead to psychological strain and stress. It was also proposed that
this perception of balance could only obtain if all relevant aspect of a
person's work environment are in balance, and finally that the perception of
balance will be different for each individual.
The operationalisation of such abstract holistic concepts as sense-making
and meaning is not an easy task, as witnessed by the literature where one
often finds only limited attempts at empirically measuring such complex
processes. My attempt started from existing materials within the psychology
literature on stress, and tried to complement those with instruments and
scales borrowed from other areas or developed for this specific purpose.
The empirical support for the theoretical model hinges of course on the
success of my operationalisation attempts. If one assumes that the
208 Steven Verjans
instruments used can measure such subjective cognitive variables as

meaningfulness and stress, then the data does indeed provide partial support
of the individual-level model. The very strong correlations between such
measures as meaningfulness, predictability and influence, seem to indicate
the existence of a general 'meaning'variable, which is strongly related to job
satisfaction and organisational commitment on the one hand, and to strain /
stress on the other. The data also shows support for the proposal that
integrity of meaning links individual psychological variables to
organisational and technology-related variables, thus indicating that
psychological well-being can only be obtained if all aspects of the individual
and his/her environment are in balance. Finally, the data supports the
importance of individual differences in perceiving one's environment. Each
person will use a different filter when viewing changes in his/her
environment, and these filters need to be considered when introducing such
organisational changes as the implementation of new technology.
At the organisational level, a similar theory was developed which shows
resemblance to work within organisational theory on contingency and multi-
contingency frameworks, frameworks that also rely on concepts of fit and
balance in determining organisational well-being. In analogy with the
individual level, concepts of organisational strain and stress were developed,
indicating situations characterised by imbalances or misfits between an
organisation and its environment or within an organisation.
The operationalisation of the model on the organisational level uses
existing instruments for gathering and analysing data, and uses both
qualitative and quantitative techniques. The data show an organisation where
there is a high level of organisational strain, due to the absence of a shared
view amongst stakeholders on crucial aspects of Omikron's current situation
and organisational design. Moreover, the data analysis showed a high
number of situational and contingency misfits, indicating a company in
crisis. Whether BlueTech software will introduce a new misfit or solve
existing ones, is not clear at this point, since there is no agreement within the
organisation regarding a strategy for information technology, partly because
there is no shared view on what the strategy and vision for Omikron are.
There does not seem to be a joint cognitive framework about the
organisation or its environment.
There are quite some limitations to the research described here, even if
one takes its exploratory nature into account. Firstly, the data that was
gathered concerned only a very small piece of software that is voluntary
used by a handful of people. This type of technology is not expected to have
a great impact on the user's perception of himself or her environment, nor on
the organisation that is implementing it. Secondly, the data gathering
occurred only after the implementation of the software and not before, so
there was no way to investigate the model in its longitudinal aspects.

Thirdly, the data are biased by the high representation of pipe shop workers,
and the sample is not very big.
In the near future, some of these limitations will be solved in the second
case study at Alpha Industries that is part of this research project, where data
were gathered about an organisational redesign and IT implementation
project that is expected to have a large impact on the participants and their
organisation. Moreover, these data have been gathered before the start of the
change project and some 3-4 months after the implementation of the change,
so a comparison through time will be possible. The Alpha industries case
study is smaller than the one described here, but it will certainly throw some
more light on the theory and its operationalisation.
REFERENCES
Arnold, J., Robertson, 1. T., & Cooper, C. L. (1995). Work Psychology: Understanding human
behaviour in the workplace (2nd ed.). London: Pitman.
Barley, S. R. (1986). Technology as an occasion for structuring: evidence from observations
of CT scanners and the social order of radiology departments. Administrative Science
Quarterly, 31 (1), 78- \08.
Burton, R. M., Lauridsen, J., & Obel, B. (2000). Fit and misfits in the multi-dimensional
contingency model: An organizational change perspective ( LOK report No. 1-2000).
Odense: University of Southern Denmark.
Burton, R. M., Lauridsen, J., & Obel, B. (Forthcoming). The opportunity cost of extreme
situational and contingency misfits.
Burton, R. M., & Obel, B. (1998). Strategic organisational diagnosis and design: Developing
theory for application ( 2nd ed.). BostonlDordrechtlLondon: Kluwer Acadmic Publishers.
Caplan, R. D. (1998). Person-Environment Fit. In J. M. Stellman (Ed.), Encyclopaedia of
Occupational Health and Safety (Vol. 1, I'p. 34.15-34.17). Geneva: International Labour
Office.
Cooper, C. L., Sloan, S., & Williams, S. (1987). Occupational Stress Indicator. Windsor:
NFERINelson.
DeSanctis, G., & Poole, M. S. (1994). Capturing the complexity in advanced technology use:
adaptive structuration theory. Organization Science, 5(2),121-147.
D0jbak, D., Nissen, T., & Obel, B. (2000). OrgCon: Vejen til at here din virksomhed bedre at
kende. Ledelse i dag, Forthcoming.
Edwards, J. R., Caplan, R. D., & Harrison, R. V. (1998). Person-environment fit theory:
conceptual foundations, empirical evidence and directions for future research. In C. L.
Cooper (Ed.), Theories of Organizational Stress (pp. 28-67). New York: Oxford University
Press.
Eisenhardt, K. M. (1989). Building theories from case study research. Academy of
Management Review, 14(4),532-550.
Feldman, M. S. (1989). Order Without Design. Stanford, CA: Stanford University Press.
Franz, C. R., & Robey, D. (1987). Strategies for research on information systems in
organizations: a critical analysis of research purpose and time frame. In R. J. Boland, Jr. &
210 Steven Verjans
R. A. Hirschheim (Eds.), Critical Issues in Information Systems Research (pp. 205-225).

Chichester: John Wiley & Sons.
Fulk,1. (1993). Social construction of communication technology. Academy of Management
Journal, 36(5), 921-950.
Giddens, A. (1979). Central Problems in Social Theory. Berkeley, CA: University of
California Press.
Giddens, A. (1984). The Constitution of Society. Cambridge: Polity Press.
Glaser, B. G., & Strauss, A. L. (1967). The Discovery of Grounded Theory: Strategies for
Qualitative Research. New York, NY: Aldine Publishing.
Hackman, J. R., & Oldham, G. R. (1975). Development of the Job Diagnostic Survey. Journal
of Applied Psychology, 60(2), 159-170.
Hackman, 1. R., & Oldham, G. R. (1980). Work redesign. Reading, MA: Addison-Wesley.
Harrison, R. V. (1978). Person-environment fit and job stress. In C. L. Cooper & R. Payne
(Eds.), Stress At Work (pp. 175-205). New York: John Wiley & Sons.
Harrison, R. V. (1985). The person-environment fit model and the study of job stress. In T. A.
Beehr & R. S. Bhagat (Eds.), Human Stress and Cognition in Organizations (pp. 23-55).
New York: John Wiley & Sons.
Hofstede, G. (1980). Culture's Consequence: International Differences in Work-related
Values. Beverly Hills, CA: Sage.
James, L. R., James, L. A., & Ashe, D. K. (1990). The meaning of organizations: The role of
cognition and values. In B. Schneider (Ed.), Organizational Climate and Culture (pp. 40-
84). San Francisco, CA: Jossey-Bass Publishers.
Kanfer, R. (1990). Motivation theory and industrial and organizational psychology. In M. D.
Dunnette & L. M. Hough (Eds.), Handbook ofIndustrial and Organizational Psychology
(Vol. I, pp. 75-170). Palo Alto, CA: Consulting Psychologists Press.
Karasek, R. (1998). Demand/control model: A social, emotional, and physiological approach
to stress risk and active behaviour development. In J. M. Stellman (Ed.), Encyclopaedia of
Occupational Health and Safety (Vol. 1, pp. 34.36-34.15). Geneva: International Labour
Office.
Kristensen, T. S. (1999). Sp0rgeskema om psykisk arbejdsmilj0. K0benhavn:
Arbejdsmilj0instituttet (National Institute of Occupational Health).
Locke, E. A. (1996). Motivation through conscious goal setting. Appplied and Preventive
Psychology, 5, 117-124.
Miles, M. B., & Huberman, A. M. (1984). Qualitative Data Analysis: A Sourcebook of New
Methods. Newbury Park, CA: Sage Publications.
Morgan, G. (1986). Images of Organization. Beverly Hills, CA: Sage Publications.
Mottaz, C. J. (1988). Determinants of organizational commitment. Human Relations, 41, 467-
482.
Orlikowski, W. J., & Gash, D. C. (1994). Technological frames: making sense of information
technology in organizations. ACM Transactions on Information Systems, 12(2), 174-207.
Orlikowski, W. 1., & Robey, D. (1991). Information technology and the structuring of
organizations. Information Systems Research, 2(2), 143-169.
Pinder, C. C. (1998). Work Motivation in Organizational Behavior. London: Prentice-Hall.
Prasad, P. (1993). Symbolic processes in the implementation of technological change: a
symbolic interactionist study of work computerization. Academy of Management Journal,
36, 1400-1429.
Schabracq, M. J. (1991). De inrichting van de werkelijkheid (The design of Reality).
MeppellAmsterdam: Boom.
Schabracq, M. J., & Cooper, C. L. (1998). Toward a phenomenological framework for the
study of work and organizational stress. Human Relations, 51 (5), 625-648.
Starbuck, W. H., & Milliken, F. J. (1988). Executives' perceptual filters: What they notice and
how they make sense. In D. C. Hambrick (Ed.), The Executive Effect: Concepts and
Methods for Studying Top Managers (pp. 35-65). Greenwich, CT: JAI Press.
Staw, B. M. (1991). Dressing up like an organization: when psychological theories can
explain organizational action. Journal of Management, 64, 273-290.
Verjans, S. (Forthcoming). Harmony and Stress in Information Systems Development and
Implementation. A Multilevel Theory and Some Empirical Work on the Crossroads of
Work Psychology, Organisational Theory and Information Systems Research - PhD
Dissertation. Odense: University of Southern Denmark Press.
Verjans, S., Mogensen, L. K., & Lynggaard, H.-J. B. (1998). Interactive collaborative
production scheduling. Proceedings of the 24th Annual Conference of the IEEE Industrial
Electronics Society - IECON'98, 1926-1931.
Weick, K. E. (1990). Technology as an equivoque: sensemaking in new technologies. In P. S.
Goodman & L. Sproull (Eds.), Technology and Organisations (pp. 1-44). San Francisco,
CA: Jossey-Bass Publishers.
Weick, K. E. (1995). Sensemaking in Organizations. Thousand Oaks, CA: Sage.
Whyte, W. F. (Ed.). (1991). Participatory Action Research (Vol. 123). Newbury Park: Sage
Publications.
Yin, R. K. (1984). Case Study Research. Beverly Hills, CA: Sage Publications.
Chapter 10
A Semiotic Approach to Improve the Design Quality

of Agent-Based Information Systems
Samuel Chong and Kecheng Liu
1. INTRODUCTION
Emerging technologies such as software agents are making information

systems easier to implement and use by allowing people to delegate work to
the computer. Software agents were first used several years ago to find and
filter information, automate routine behaviour and match people with similar
interests. Open environments such as the Internet and World Wide Web have
also enabled the increasing use of software agents in information systems.
Today the field of software agency is a promising research which brings
together and draws on results from many disciplines, such as artificial
intelligence (AI), computer science, sociology, economics, organisational
behaviour and philosophy. Multi-agent systems consist of several software
agents that cooperate to perform a common task. Because of their wide
scope and multi-disciplinary nature, it is difficult to define software agents in
a few words. The term "agent" has been so widely used that, according to
Finin et al. (1997), anything might qualify as an agent these days. To serve
as a starting point and as a constant point of reference for reading this paper,
we define software agents as:
"autonomous software entities having the abilities to sense on and react
to their environment, as well as to communicate and cooperate with other
214 Samuel Chong and Kecheng Liu
agents in order to accomplish their duties or functions which are delegated

from their human users".
It is worthwhile to emphasise that while the human users merely delegate
their duties to the software agents, the software agents cannot be held
responsible for their actions. As multi-agent systems get more sophisticated,
the role of software agents cannot be seen as just some instrumental means
of production. Software agents take a very active part in the human
communication and co-ordination process. However, the 'intelligence of
software agents has a completely different notion from human's as they are
always programmed by humans, however sophisticated software agents are.
Although in this paper, sometimes the distinction between a human and
software agent is not made explicitly, these two types of agents are totally
different in their nature in terms of responsibility they can take. In designing
agent-based systems, it is imperative to have a mechanism for tracing the
responsibility. Therefore new requirements modelling methods are necessary
to address this very issue. In recent years there has been a growing interest in
requirements modelling (Davis 1993, Loucopoulos & Karakostas 1995,
Macualay 1995, Wieringa 1996). However, existing requirements modelling
methods such as the objected oriented analysis and design (Booch 1994,
Coleman et al 1994), UML (Fowler & Scott 1999) and the corresponding
CASE-tools (Boggs & Boggs 1999) only partially fit the needs of modelling
multi-agent systems. They fail to adequately model an agent's autonomous
behaviour and the richness of an agent's interaction. The reasons lie in the
fundamental difference between an object and an agent. A key concept of
agent is that of autonomy, where it is aimed at constructing more
"intelligent" software, which not only reacts to its user (reactive), but also
anticipates its user's needs and interests (proactive) (Maes 1995). Due to its
"intelligence", software agents have the ability to change a social state of its
user (e.g. make a decision to make a credit card payment on behalf of its
user). Design methods that are appropriate for technical systems cannot
simply be applied to a system that is viewed as a part of the human
communication and social system. Friedman & Kahn Jr. (1994) gave two
examples of computer systems that are competent in the technical aspects,
but made little sense for the social context of their use. Existing methods
have generally ignored the importance of the human and social issues in the
requirements modelling process and lack any rigor or formal basis upon
which to model the social aspects of agents. Organisations are by nature
complex social systems. In essence, what is needed is a social approach that
is capable of capturing the social aspects of the organisation to be supported
by a multi-agent system rather than merely the functions the multi-agent
system needs to perform. The social approach is one in which a multi-agent
system is viewed as part of the whole social system by placing it within the
A Semiotic Approach to Improve the Design Quality... 215
broad organisational environment with the user as an integral part of the

system.
The aim of our work is to develop a methodology for modelling the
requirements of software agents using a set of semiotic methods. The
semiotic methods offer a penetrating insight into the human and social
aspects of multi-agent systems neglected in today's method for analysis and
design. Organisational semiotics, upon which our work is based on, is a field
of study that observes the use of signs and languages in organisations
(Stamper 1973, Stamper et al. 2000). In order to better understand agents in
organisations, a model is constructed through the observation of the use of
signs or languages in the organisation. The organisation model (with a
semantic model at its core) includes all the meanings and the conceptual
dependencies of terms, roles of agents and their patterns of behaviour. The
methodology also makes a distinction by encompassing a mechanism and a
standard notation for specifying the mental states and prescriptive behaviour
of agents by eliciting norms. Therefore, the focus of our research work is on
the modelling of complex business requirements, not only by establishing a
means of integrating different information sources, but also by establishing a
means to create a stable environment in which software agents work by
identifying the norms that serve to govern their patterns of behaviour.
2. ORGANISATIONAL SEMIOTICS
.
~ Sociel. world- belief., expectations.
Human commitments, contracts,law. culture•......
Information Pragmatic,' - in1enti.o.n~ commwUcation,
Functions . conve~attons. neg~~II!lt.ions•. "." ..
-------------------Ts~;;tic.-~~1~~f:~~~~I~;~~;:~~~~;d.~-----------------
The IT PlatformI deduction, ,o!\wore. fil ........ .
J
Empiric.- pattell\, variety. noin, entropy, channel capacity, rectu.ndancy.
1------' efficiency. codes._ .. ...
Physical worlcl- Signall, tracI', phylic&t di.tinction•. hardware, co mponent density.
soeed. economic. . .....
Figure 1. The semiotic framework
Traditionally, there are three distinct fields of semiotics, which are

syntactics, semantics and pragmatics, which deal with the combination of
signs, uses of signs to communicate meanings and uses of signs to
communicate intentions respectively. These three fields reflect the
philosophical foundation of semiotics. To these three headings of semiotics,
Stamper (1973) added three more. The introductions are the physics of signs,
empirics of signs and social world level of signs. They deal with the physical
aspects of signs, statistical properties of signs and the social effects of the
use of signs. Adding these three extra aspects of signs, he arrived at a fuller
semiotic framework as shown in figure 1. These introductions later form the
basis for developing a set of semiotic methods for studying the use of signs
in an organisation, a sub-field of what is now better known as organisational
semiotics.
Organisational semiotics when applied to multi-agent systems
development opens another promising perspective. Andersen (1997), Clarke
(1998) and Liu (2000) each presents an illustration of developing
information systems based on such a paradigm. It is important to note that
though the sign is the fundamental notion in organisational semiotics, it is
not required that all approaches based on it to use the concept of signs alone
(see Clarke (1998) for an example). In our work, a multi-agent system based
on this paradigm is viewed as a part of the social and human communication
systems. The functionalities of such an information system are directly
established by the business and social requirements of the organisation,
which can be manifested through the use of signs. This is illustrated in figure
2. The uses of signs and languages playa central role in the functioning of
such an information system because signs of one kind of another are created
and exploited by software agents to accomplish the tasks delegated by their
human users. In order to understand and computerise the signs that are
created and used in a society of agents, we first need to study the signs that
are created and used by the human actors. In general, we are interested in
studying how meanings are conveyed by natural language and what aspects
of language need to be captured through the representation of data so that the
multi-agent system can function as an effective substitute for natural
communication. The language to be studied covers not only languages of
different countries, but also the technical vocabularies of different
professions and organisations.
Social and human communication system
Agent. based information system
Figure 2. Multi-agent system as part of the social system

The semiotic framework leads us to examine different aspects of multi-

agent systems in terms of the properties of signs on different layers. Since
we have advocated earlier in this paper that multi-agent systems are a part of
the human communication system, we are particularly interested in the
human information functions levels. In the next section, the effects of the use
of signs by a society of agents is analysed using the information functions
level.
2.1 Social Effect of the Use of Signs by Software Agents
Communication acts are the means by which agents use language to

achieve their goals. However, before any effective communication can take
place, agents must first understand the terms that each other use. At the
semantic layer, signs are used to communicate meanings. This aspect of
signs is essential for effective agent-to-agent communication. This
corresponds to the need for a shared meaning of terms, most commonly
referred to as an 'ontology' in artificial intelligence. Explicitly defining the
meaning of every concept and term is tedious, hence the work on tools and
editors such as Ontolingua (Gruber 1991. 1993) and ZEUS (Collis et al.
1998). However, existing methodologies or tools capture the terms on an ad
hoc basis (e.g. Taveter 1999), without a formal basis in which the terms and
their meanings can be represented explicitly and systematically, as is the
purpose of . the methodology proposed here. At the pragmatic layer, by
using signs that have shared meaning, agents can communicate their
intentions to other agents. Using the mentalistic metaphor, intentions are part
of a software agent's 'mental state'. Generally, in order for these agents to be
proactive and predict the possible behaviour of other agents in their
environment, it is essential for agents to maintain knowledge about their
environment. The knowledge may include other agents' patterns of
behaviour, intentions and beliefs, which create and maintain the order and
stability in the social context in which the agents operate by creating
expectations about the behaviour and mental states of other agents. It is
essential that the methodology be able to capture the mental states of agents
in order to create this stability.
At the social world layer, agents have the ability to change the social
state of its user once they have the ability unambiguously to communicate
their meanings and intentions. For example, an agent expressing an intention
as a promise to buy an item will later incur a bill on its user's credit card
account. The methodology must be robust enough to identify the possible
patterns of behaviour and associate constraints on the patterns of behaviour.
For example, an agent having the "buy" function will have a constraint
associated to it, such as "permitted to make payment only if the price is less
than $100". At the same time, it should be able to manifest the responsible
human users, so that analysts can verify the constraints with those
responsible human users. The proposed methodology, will attempt to handle
these issues at the social world layer by identifying the patterns of behaviour
of agents and the responsible human users in a graphical form and by
formalising the constraints in the form of norms using a standard notation.
The proposed methodology uses two methods, namely, the Semantic
Analysis and Norm Analysis methods. They are an attempt to answer the
following questions:
1. How do we identify the terms that are to be included in the analysis?
2. How do we identify the types and roles of agents?
3. How do we identify the mental states of agents?
4. How do we distinguish the responsible human actors?
5. How to understand how multiple agents should behave in order not to
infringe the rights oftheir owners?
3. SEMANTIC ANALYSIS
Semantic analysis (Stamper 1979, 1985) enables one to understand the

business domain through a rigorous process that reveals the dependencies of
concepts upon one another clearly expressing them in a graphical form
known as the 'semantic model'. For the sake of brevity, we will not be going
into details regarding the steps on obtaining the semantic model. Readers
may consult Liu (2000) for more details. The semantic model in figure 3 is
based on a case study carried out on an online insurance company. It
identifies three main concepts, which are actors (in circles), patterns of
behaviour (in nodes) and roles (in parentheses).
The first strength of the semantic model is that it establishes the core of
the meanings of the terms in use within the organisation. For example, the
information-rich semantic model in figure 3 reveals many terms used in the
application domain, such as "policies", "QQ", "subscribes" and so on. The
communication between the different human users in the organisation cannot
be effective without a hared understanding of these terms. Likewise, the
software agents that work on behalf of the human users, need to be able to
access the meaning of these terms before their communications can gain any
value.
Secondly, the model provides a conceptual framework for identifying the
roles and positions of agents (in parentheses) within the organisation and the
responsible human users or organisations (in ovals) who have the right to
delegate their duties to the agents. Therefore, the model serves also as a
communication medium in which the analysts can engage the responsible
human users in a dialectic exploration of the mental states of their agents and
the constraints or rights to be placed on the agents' patterns of behaviour
(such as "buys"). The semantic model provides not only information about
the business domain in which the agents have to work, but also a context for
understanding the current policies of the organisation. Because analysts
know which responsible human user to approach, miscommunication of the
requirements can be avoided as the true solution for the true problem can be
developed and stated. After all, if there is anything that distinguishes human
from animals, it is human's ability to talk and clarify any doubts that may
arise in the process. The mental states and constraints or rights are expressed
in the form of norms, which will be discussed in the next section.
Figure 3. Semantic model ofan online insurance company.
4. NORM ANALYSIS
Stamper argues that an organisation has many layers of information

systems among which are the three main layers shown in figure 4:
The informal information system (IS) comprises the informal use of signs
and languages in organisations to get things done. At this layer, tacit
knowledge constitutes part of the behaviour of the members of the
organisation and enable them to understand what each other means and
intends without any conscious effort. Much of the tacit knowledge is learned,
passed on and applied without applying an explicit level of thought. When
the tasks to be performed and the communication between members
becomes fairly repetitive, the rules for performing the tasks and
communication may be formalised to enable the efficient execution of those

tasks and the communication between members. This gives rise to the
bureaucratic or formal IS. The formal IS is dependent on the explicitly
formalised rules which govern the mechanical handling and transmitting of
signs. Eventually, the need to automate part of the formal IS gives rise to yet
another layer of information system, the IT system. In the IT system, parts of
the formal IS are automated by having computers to read signs without
understanding them and to manipulate signs according to rules.
Informal IS
Formal IS
IT system
Figure 4. Three layers of organisational formality
A key unifying concept that links these three layers together is the
concept of norms. Norms exist in all three layers of the information systems
in an organisation. However it is at the informal IS layer that matters most in
our approach when modelling the requirements of a multi-agent system.
Therefore we are not particularly concerned about the explicitly formalised
norms in the formal and IT level. Members of an organisation may still
behave in a coordinated way even when there are no explicit rules being laid
down (e.g. distribution of duties and responsibilities informally among
colleagues). It is at the informal IS level where norms determine the mental
states and govern the behaviour of the members of an organisation without
their being consciously aware of it.
The study of norms had been a growing concern in the AI field (see
Moses & Tennenholtz 1992; Shoham & Tennenholtz 1992). Norms help
members of an organisation to establish what patterns of behaviour are legal
and acceptable within a given social context. An individual member in the
organisation, having learned the norms, will be able to use the knowledge to
guide his or her actions. Four types of norms can be identified. All of them,
in one way or another, affect a person's intentions and actions. They are
evaluative norms, perceptual norms, cognitive norms and behavioural norms
(Liu & Dix 1997; Chong & Liu 1999, Chong & Liu 2000, Stamper et al.
(2000). Each type of norms governs human behaviour from different aspects.
Perceptual norms are concerned with how people perceive patterns in the
signals from the environment such as sight, sound and taste they receive
through their senses.
Each of these norms manifests either a mental state or a pattern of
behaviour that we adopt. Perceptual norms are concern with the way in
which we divide up the world into the phenomena to which we attach names
such as marriages, poverty and copyright. Evaluative norms allow us to
make judgements about what we have felt and recognised and, indeed to
decide what perceived patterns are worthy of repeated recognition.
Encountering something judged bad usually creates an intention to escape it.
Cognitive norms can be recognised because their consequent parts affect our
beliefs respectively. For example, it is common to find that people have false
beliefs about the reliability of the actual deliveries of the products bought
over the Internet. This general belief may affect the intentions of customers.
As a result, customers may not intend to buy any products through the
Internet unless they are very certain about the reliability of the actual
deliveries. Behavioural norms govern how people or organisations behave.
Using the previous example, in order to gain the trust of the customers,
behavioural norms may state that organisations are obliged to deliver
products on the agreed date, otherwise a penalty will be imposed. Any
intention can be treated as a directive given to oneself to do or not to do
something.
4.1 Representing Behavioural Norms
Behavioural norms are more observable and are the ones that affect and
regulate humans' behaviour in an organisation. They have a prescriptive and
proscriptive function in governing the behaviour of agents and are expressed
in the form of "you are obliged, permitted or forbidden to behave in certain
way". In general, behavioural norms have the following structure:
IF <certain conditions apply>
THEN <agent>
is <obliged/permitted/forbidden>
To perform <action/speech act> or adopt <a state>
Textbox 1 shows some of the captured behavioural norms that control the
process of the buying of insurance on the Internet. They are elicited by
examining each of the patterns of behaviour in the semantic model in figure
3.
Nann: N1
N1: IF a person has an agency contract with an insurance company
THEN in the role of insurance agent that person
is pennitted
to state a quotation to a client
Nann: N2
IF buyer agent wishes to buy a policy and the price is less than $100
THEN buyer agent
is permitted
to make payment using the human user's credit card number
Nonn: N3
IF buyer has accepted quotation
THEN insurance agent
is obliged
to sell policy at the quoted price
Nonn: N4
IF buyer has accepted quotation
THEN buyer
is obliged
to buy policy at the agreed price
Textbox 1. Behavioural norms.
Behavioural norms provide the mechanism in which the constraints or the

rights on the agents' patterns of behaviour can be captured. The patterns of
behaviour identified in the semantic model are defined at very abstract level.
It does not show the constraints or the rights of agents. In norm analysis, one
of the explicit focuses is given to capturing the behaviour of an agent in the
form of "should", "permitted" and "forbidden". In this process, the
responsible human users (identified from the semantic model) are consulted.
The relationship between behaviour norms and the patterns of behaviour is
illustrated in figure 5.
Figure 5. Relationship between pattern of behaviour and behavioural norm

The particular norms delegated to the software agent cover not only its
own permissions and obligations but also those of other participants such as
the insurance agent and the credit card company. In additions the agent
would be given the relevant norms of commercial law. Moreover it would
make use of the semantic model, which contains the basic perceptual norms
for the application domain. Using these norms, the software agent can form a
model of the state of affairs in which it is operating.
Representing behavioural norms in any multi-agent system formally
requires a standard notation. As a basis for a formalism we propose to
examine Deontic Logic (von Wright 1968; Myer & Wieringa 1993). The
purpose of the notation is to represent the behavioural norms designed to
regulate the behaviour of agents in multi-agent systems. It also serves as a
standard mechanism for specifying behavioural norms and thus enhances
clarity and co-operation between analysts and developers. In order to specify
the intended set of behaviours related with an action or speech act, the three
fundamental deontic operators are introduced in the notation. They are
permitted, obliged and forbidden. A permitted action or speech act is one
which can be performed legally by an agent, an obligatory action or speech
act is one which must eventually be performed by an agent if the obligation
is to end and a forbidden action or speech act is one which cannot be
preformed by an agent.
A part of the structure of the proposed notation is expressed in terms of
first order logic operators such as: -, (Negation), A (And), V (Or) and of the
quantifiers V (for all) and :J (there exists). The reader may consult a
textbook, such as Galton (1993) for details. The three, deontic operators are
expressed in P (Permitted), 0 (Obliged) and F (Forbidden). The notation
consists of the following axioms and rules (e.g. the construct 0 (p/q) is read
as p is obligatory under condition q):
(AI) 0 (p A q/r) ~ 0 (p/r) A 0 (q/r)

(A2) 0 (p/q V r) ~ 0 (p/q) V 0 (p/r)
(A3) 0 (p/q) ~ P (p/q) (Obligatory to do p under condition q if and only
if it is permitted to do p under condition q).
(A4) P (p/q) ~ -, F (p/q) (Permitted to do p under condition q if and
only if it is not forbidden to do p under condition q).
(AS) P (p/q) ~ -,0 (-,p/q) (Permission is the dual of obligation).
(A6) 0 (p/q) ~ -,P (-,p/q) (Obligatory to do p under condition q if and
only if not permitted not to do p under condition q).
(A 7) -, (0 (p/q) A 0 (-,p/q) ) (Obligatory to do p under condition q and
obligatory not to do p under condition q is contradictory).
For example, the norm NI in textbox 1 will be formulated as follows:
Nl. Vx Pa (insurance agent(x) ) == (~ (x) )

where a = state a quotation to clients and ~ = registered with insurance
company
The symbol a and ~ are introduced as actions or speech acts variables.

They can be an action or a speech act. Nl is interpreted as it is permitted for
all x, where x is an insurance agent, to state a quotation to clients if x is
registered with insurance company. According to the axiom A3, an
equivalent formulation could be:
Nl. Vx -.Fa (insurance agent(x) ) == (~ (x) )
where a = state a quotation to clients and ~ = registered with insurance

company.
It is not forbidden for all x, where x is an insurance agent, to state a
quotation to clients if x is registered with insurance company.
N2, N3 and N4 are formulated as follows:
N2. Vx P~ (buyer (x) ) == (a (x) )

Where ~ = make payment using credit card and a = wishes to buy a
policy and the price is less than $100.
It is Permitted for all x, where x is an insurance agent, to make payment
using credit card if x wishes to buy a policy and the price is less than
$100.
N3. Vx Oa (insurance agent (x) ) == ( (::Jy) ~ (buyer (y) )

Where a = sell the policy at the quoted price and ~ = accepted the
quotation.
It is obligatory for all x, where x is an insurance agent, to sell the policy
at the quoted price if there exist some y, where y is a buyer agent, who has
accepted the quotation.
N4. Vx Oa (buyer (x) ) == (a (x) )

where a = buy the policy at the agreed price and ~ = accepted the
quotation.
It is obligatory for all x, where x is a buyer agent, to buy the policy at the
agreed price if x had accepted the quotation.
4.2 Representing Other Norms
Perceptual, evaluative, cognitive and intentional norms are also the basis
of organised behaviour. All norms can serve a descriptive function in
depicting the mental states of individual members. For example, the ability
to recognise the term "computer" and believe that it is capable of performing
a word processing operation and having an intention of using it to perform a
word processing operation. The consequent parts of the norms reflect the
mental states of each member.
All norms have the following structure:
IF <certain conditions apply>
THEN <agent>
Adopts <an attitude>
Towards < some thing or proposition>
Textbox 2 shows some of the intrinsic norms that were captured by

examining each of the terms and patterns of behaviour in the semantic
model.
Norm: N5
IF buyer wishes to buy an insurance policy.
THEN buyer
will N'ltentto
search for that policy
Or buyer
will intent to
ask for a quotation through e-mail
Norm: N6
IF buyer had enquired about a policy.
THEN buyer
will eJqject
to get a quotation
Norm: N7
IF buyer has bought a policy.
THEN buyer
will believe
that its owner is covered under that policy
Norm: N8
IF insurance wishes to subscribes to QQ.
THEN insurance agent
will acknowlectJe
it has to pay a $100 subscription fee.
Textbox 2. Intrinsic norms

The semantic model in figure 3 includes many terms . The semantic

model is a result of respecting the viewpoints of different social groups and
deciding which meaning to give to a term in that social group. The meanings
of terms can only be established by considering the perceptual norms or
evaluative norms of the human users. The strength of norm analysis is that it
provides a mechanism for eliciting the meanings of the terms that were
already identified in the semantic model. The relationship between the terms
and norms can be shown in the illustrations in figure 6.
pe""Ir_ _ _ ~C2\~_b....:Uyr_r~_ _ bUYS _,- -,--,: - p;~;;~:;~; E.-Q~~;;; --- ----.

./ , .' : Norm : • .g. IF buy .. wilh .. la
/ _---' .- ' - : buy policy. Then buyer
~ _ _ po!icit, .~:____ : acmowledge'pohcycon,isloof
~ .~------------J po6cy_lype.policy_no. etc __ _
e - .. .' :
INlInlICII! e~n ~- T-;.~~~;~~I ~~ -g;;~~~.;~ --------
.-- : Norm : e.g. IF insurance .gent
wi,hulooublCribeloQQ,
.,." - '* : TH EN insurance aged
QQ IWscnbe., : : __ . : .cknowledg... ubscribe·paying
--------------_L $100
Figure 6. Relationship between terms and perceptual or evaluative norms
Similarly, norm analysis also allows an analyst to elicit the intentional

and cognitive norms of the human users. This is essential in a successful
implementation of a multi-agent system as it helps agents to predict the
intentions and beliefs of other agents and thus creates a stable environment
in which the agents work in. The analyst begins by examining each pattern
of behaviour and identifies the intentional and cognitive norms that are
associated with it. The responsible human users are consulted in this process.
The relationship between intentional and cognitive norms and patterns of
behaviour can be illustrated in the figure 7.
.--. T;';;;~;;1-;;
· C;;I~;":- --.
: No",,: If buyer ",i,h •• to
: buy polity. TH EN buyer
: will intend to enqui,re
.. . :- _____III:iIIl.
: throultl e- mail
Figure 7. Relationship between pattern of behaviour and intentional or cognitive norm
First order logic operators are also use to represent intrinsic norms, with
the exclusion of the three deontic operators. Instead, the Int (Intention), Bel
(Belief), Per (Perception) and Val (Values) operators are introduced. The a
and ~ variables are used to represent the "conditions" and "consequences"
part of the intrinsic norms. However, first order logical statements can only
refer to one state at a time. This is insufficient for expressing the evolution of
the "attitudes" of agents. We try to overcome this problem by offering an
additional temporal operator "0". Sub-statements are prefixed by an 0
operator and are interpreted as in the next state. The new operator and its
intuitive meaning is given below (<1> is a consequent statement):
0<1> I <1> holds in the next state
The notation for representing intrinsic norms consists of the following

axioms and rules:
(A I) P ~ 0 <1> (p (condition) implies (/J (consequent) )

(A2) 0 (<1> A '1') ~ (0 <1» A (0 '1')
(A3) 0 (<1> V '1') ~ (0 <1» V (0 '1')
(A4) -, 0 (0 <1> A -,0 <1» (There cannot be two contradictory consequent)
(AS) 0 <1> ~ -, (-,0<1»
(Note: (/J and Ij/ are consequent statements)
For example, the norm NS in textbox 2 is expressed as:
'r/x a (Buyer (x) ) ~ 0 (intent (x, ~) V intent (x, ~') ) / Int

where a = wishes to buy an insurance policy, ~ = search for that policy
and ~' = ask for a quotation through e-mail
(note: (/J / Int is read as having the disposition of intention under the
statement C/J)
It is interpreted as for all x who wishes to buy an insurance policy, where

x is a buyer, implies that x will intent to search for that policy or ask for a
quotation through e-mail / an intention
According to axiom A3, N5 could alternatively be expressed as:
Vx a (Buyer (x) ) ~ 0 (intent (x, 13 V 13') ) / Int

N6. Vx a (Buyer (x) ) ~ 0 (expect (x, (3» / Bel
where a = had enquired about a policy and 13 = get a quotation
For all x who had enquired about a policy, where x is a buyer, implies
that x will believe it will get a quotation / a belief
N7. Vx a (Buyer (x» ~ 0 (belief (x, (3» / Bel

where a = bought a policy and 13 = owner is covered under that policy
For all x who bought a policy, where x is a buyer, implies that x will
believe that its owner is covered under that policy / a belief
N8. Vx a (insurance agent (x) ) ~ 0 (acknowledge (x, (3» / Per

where a = wishes to subscribes to QQ and 13 = has to pay $100
subscription fee
For all x who wishes to subscribe to QQ, where x is an insurance agent,
implies that x acknowledges that x has to pay $100 subscription fee.
By standardising a mechanism for formulating behavioural norms and

intrinsic norms, analysts are forced to speak the same language. Co-
operation between analysts and developers can be enhanced as there is a
formal ground for shared understanding.
The norms captured and represented above can be grouped into a role
schema and term schema, where explicit links are made between norms and
the patterns of behaviour or terms as shown in figure 8 and 9.
&le: buyer Pattern of behaviour: buys
Constraints/rights:
'ltxPI3 (buyer (x) '" (0'. (x»
Where 13 '" make payment using credit card and 0'. '" wishes to buy a policy and Behavioural
the price is less than $100. norms
'ltx 00'. (buyer (x) ) l!!! (0'. (x) )
where 0'. '" buy the policy at the agreed price and 13 '" accepted the quotation.
I~ Intentional
MlInta1 statlls: or cognitive
'ltx 0'. (Buyer (x) ) ~ <> (intent (x, (3) V intent (x, 13') ) lInt
norms
where 0'. '" wishes to buy an insurance policy, 13 '" search for that policy and 13' '"
ask for a quotation through e-mail
'ltx 0'. (Buyer (x) ) ~ <> (belief (x, 13» I Bel

where 0'. '" bought a policy and 13 '" owner is covered under that policy
I....
Figure 8. Role schema
Term: subscribes
Meaning: N8. Ifll; a. (insurance agent (x) ) ~ 0 (acknowledge (x, 13» I Per }- Perceptual
where a. = wishes to subscribes to QQ and 13 = has to pay $100 subscription fee
or evaluative
~ _ _- - - - - . J norms
Figure 9. Term schema
A role schema serves to draw together all the various concepts discussed
into a single place. It comprises of the constraints or rights on the patterns of
behaviour and the mental states of agents. The role schema in figure 9
indicates that the software agent buyer has permission, as well as an
obligation, to perform some action. In addition, the agent buyer has an
intention and a belief. The term schema on the other hand, brings together
the social meaning that the human users give to a term that were already
identified in the semantic model.
5. RELATED WORK
As a result of the development and application of agent technologies in

recent years, there has been many requirements modelling methodologies for
multi-agent system. One related work is presented in Wooldridge et al.
(1999). It presents a methodology for agent-oriented analysis and design.
The methodology is founded on the view of a system as a computational

organisation consisting of various interacting roles. Though the methodology
emphasises the identification of key roles in the system, subsequently linked
to various attributes such as responsibilities and permissions, it does not
offer a formal basis in which the roles can be captured. The main
contribution of our approach distinct from the work by Wooldridge et al. is
that the semantic model provides a basis where the roles and positions of
agents in the system can be captured in a precise manner.
Another methodology is presented in Taveter (1999). The proposed
methodology captures business rules as a law or custom that guides and
constraints the behaviour of agents. Our approach is similar in that we
capture norms for the same purpose. One main contribution distinguishing
our work is that Taveter's places no emphasis on the mechanism for
identifying the terms or their semantics. The analysis in our approach
produces a semantic model which refers to a shared understanding of some
domain of interest and provides a unifying framework for capturing the
terms used and their meanings in the application area.
The work in Kinny et al (1996). propose a methodology for modelling
BOI (Belief, Desire and Intention) agents at two main levels (internal and
external). The external level viewpoint consists of decomposing the system
into agents and the definition of their interaction. The internal viewpoint
models each agent through three models, the belief model, the goal model
and the plan model. In the belief model, first order logic predicates are use
for specifying the agents' knowledge about their environment. However, it
does not include an adequate means of dealing with responsibilities. They
failed to ensure that the responsible human users are in the picture. We could
of course introduce additional predicates applied to propositions, that specify
that someone knows (or believes, thinks, expects, .. ) that some proposition is
the case, but that will ruin the mathematical elegance of first order logic.
This problem does not appear in our approach. The cognitive norms and
evaluative norms are elicited based on the results of carefully examining the
patterns of behaviour in the semantic model, with clear details of the
responsible human users.
6. CONCLUSION AND FUTURE WORK
The semantic model developed in the semantic analysis phase serves as a

unifying framework for analysing the requirements of a multi-agent system.
The semantic model identifies the list of terms and the roles of agents that
are needed to solve the problems in the business domain. Based on the
information from the semantic model, analysts can understand which human
user to rely on for any verification, such as the level of constraints or rights
to be given to an agent. The results of eliciting norms also allow the analyst
to understand how an agent should behave and what 'mental states' they
should have, which allow agents to predict the patterns of behaviour and
mental states other agents will have. In addition, norms also facilitate the
social meanings of terms to be established.
Our approach offers a systematic and insightful way of analysing the
requirements of multi-agent systems. However, much research is still needed
to further our approach to commercial exploitability. Firstly, it is important
to model agent interactions on a system wide level. Further work will
incorporate an interaction model into our approach. We envisage that the
interaction model will depict the types of interactions and the
communication acts (or speech acts) that will occur in a multi-agent system.
Norms can be further employed as social objects that ensures that all speech
acts that take place are felicitous or appropriate in a context.
Lastly, further verification and validation of our approach, which is the
main focus of our future work, needs to be done. We feel that our approach
can help analysts to better understand multi-agent system from a human and
social perspective and to design such systems to work effectively.
REFERENCES
Andersen P. B. 1997, A Theory of Computer Semiotics, Cambridge University Press, USA.
Booch, G. 1994, Objected-Oriented Analysis and Design second edition, Addison-Wesley,
Reading, MA.
Boggs, W. & Boggs, M. 1999, Mastering UML with Rational Rose, Sybex
Chong, S. & Liu, K. 1999, A Semiotic Approach To The Design of Agent-Mediated E-
Commerce Systems, Proceedings of IFIP WG 8.1 International Working Conference
ISCO-4.
Chong, S. & Liu, K. 2000, A Semiotic Approach for Modelling and Designing Agent-Based
Information Systems based on Roles and Norms, Proceedings of the International Bi-
Conference workshop on Agent-Oriented Information Systems 00' AOISOO, Stockholm
Sweden, Texas USA, to appear,.
Clarke, R. J.1998, Systemic Semiotic Development using SFX: Prototyping Genre-based
Representation of IS Workpractices, Proceedings of the 1998 IEEE ISIC/CIRNISAS Joint
Conference, Gaithersburg, MD, USA, pp. 846-849
Collis, J. c., Ndumu, D. T., Nwana, H.S. & Lee. L.C. 1998, The ZEUS Agent Building
Toolkit, BT Technology Journal, 163, pp. 60-68.
Coleman, D., Arnold, P., Bodoff, S., Dollin, C., Gilchrist, H., Hayes, F. & Jerernaes, P. 1994,
Objected-Oriented Development: The Fusion Method, Prentice Hall International, Hemel
Hempstead, UK.
Davis, A.M. 1993, Software Requirements: Objects, Functions, States, Prentice Hall.
Fowler, M. & Scott, K. 1999, UML Distilled, Second Edition: A Brief Guide to the Standard
Object Modeling Language, Addison-Wesley Pub Co
Galton, A. 1990, Logic for Information Technology, John Wiley & Sons, England.
Gruber 1991, Ontoligua: A Mechanism to Support Portable Ontologies, KSL-99-66, Stanford
University Knowledge Systems Laboratory.
Gruber 1993, A Translation Approach to Portable Ontology Specifications, Knowledge
Acquisition, 5 2, pp. 199-220.
Kinny, D., Georgeff, M. & Rao, A. 1996, A Methodology and Modelling Technique for
Systems ofBDI agents, in van der Vel de, W. & Perram, J. Eds., Agents Breaking Away:
Proceedings of the Seventh European Workshop on Modelling Autonomous Agents in a
Multi-Agent World MAAMA W'96, Lecture Notes on Artificial Intelligence Volume 1038.
Springer-Verlag: HeIdelberg, Germany.
Liu, K. & Dix, A. 1997, Norm Govemed Agents in CSCW, Proceedings of the 1st
International Workshop on Computational Semiotics, De Vinche University, Paris.
Liu, K. 2000, Semiotics in Information Systems Development, Cambridge University Press,
Cambridge, UK.
Loucopolos, P. & Karakostas 1995, Systems Requirements Engineering, McGraw Hill.
Macaulay, L. Requirements Engineering:, Springer.
Maes, P. 1995, Intelligent Software, Scientific American 273 3, Special Issue on Key
Technologies for the 21 st Century, pp. 66-68.
Meyer, J. C. & Wieringa, R. J. eds., 1993, Deontic Logic in Computer Science, Wiley,
Chichester, UK.
Moses, Y. & Tennenholtz, M. 1992, On Computational Aspects of Artificial Social Systems,
Technical Report CS91-01, Weizmann Institute.
Shoham, Y. & Tennenholtz, M. 1992, On the Synthesis of Useful Social Laws in Artificial
Societies, Proceedings of the 10th National Conference on Artificial Intelligence,
Kaufmann, pp. 276-282.
Stamper, R. K. 1973, Information in Business and Administrative Systems, 2nd edition, Basil
Blackwell, Oxford.
Stamper, R.K., 1979, Towards a Semantic Normal Form, in Database Architecture, G.
Bracchi, G.M. Nijssen, North Holland, Amsterdam
Stamper, R.K., 1985, Knowledge as Action: A Logic of Social Norms and Individual
Affordances, in Social Action and Artificial Intelligence, N. Gilbert, G. Heath eds., Gower
Press
Stamper, R., Liu, K., Hafkamp, M. & Ades, Y. 2000, Understanding the Roles of Signs and
Norms in Organisations - A Semiotic Approach to Information Design, Journal of
Behaviour and Information Technology.
Taveter, K. 1999, Business Rules' Approach to the Modelling, Design, and Implementation of
Agent-Oriented Information Systems, Proceedings of the International Bi-Conference
workshop on Agent-Oriented Information Systems 99' AOIS99, Seattle USA, Heidelberg
Germany,.
Von Wright, G. H. 1968, An Essay in Deontic Logic and the General Theory of Action, in
Acta Philosophica Fennica, vol. 21, North Amsterdam.
Wieringa, R.1. 1995, Requirements Engineering: Frameworks for Understanding, Wiley.
Wooldridge, M., Jennings, N.R. & Kinny, D. 1999, A Methodology for Agent-Oriented
Analysis and Design, Proceedings of the Third Annual Conference on Autonomous
Agents, Seattle, USA
Chapter 11
A Semiotics Approach to Analyzing the Information
Bearing Capacity of a Data Schema
Junkang Feng
1. INTRODUCTION
Many writers in the information systems area accept that an information

system is there for providing information, for example, Avison and Shah
(1997), and Flynn (1998). And yet what a computer handles is data, that is,
the systems that are normally called 'information systems' are only data
systems (Wilson 1990, Flynn 1998). So an interesting question is how data
tum into information by means of an information system. And an
information systems designer should consider what determines the capability
of information provision of a machine, which can only handle data. This
problem does not seem to have been well addressed in the literature. To
tackle this problem systematically in this chapter we will propose a notion of
'information bearing capacity of a conceptual data schema' and give a
method for analyzing such a capacity of a data schema in the Entity-
Relationship model (ER). The main significance of this work is to ascertain
that a sufficient data structure is in place for an information system to
provide required information. The chapter is organized as follows. First of
all, some related work IS examined. Then the basic concepts of this work are
presented, and the notion defined. The major part of the chapter examines
the relationship between topological connections and semantic relations with
regard to objects, and how the structure and constraints at the topological
234 Junkang Feng
level, and rules and other dependencies at the semantic level determine this
capacity in question.
2. RELATED WORK
The literature does not seem to have provided adequate knowledge on

how to approach this 'information provision' problem of an information
system, particularly how to define this notion clearly, and how to measure
this capacity. No definition for this capacity has been found in the literature.
A related topic is so called 'information content' (Batini et al 1992). Even
that is only covered by notably few writers. Eick and Lockemann (1985,
p.88) define the concept of information preserving transformation between
'S-diagrams' (a variant of the binary relation model) as whether a correct S-
diagram can be transformed into another correct S-diagram. Batini et al
(1992) suggest that a means for comparing the information content of two
schemas is to compare their ability to reply to queries. Moody (1998)
presents a set of metrics for evaluating the quality of Entity-Relationship
schemas, which does not explicitly cover the information content of such a
schema. The most relevant seems the 'completeness'. He defines four types
of completeness errors, which seem neither formal nor quantitative. None of
these works seems an adequate solution to the problem in question.
We take a semiotic approach to addressing this problem. The related
work regarding semiotics includes Stamper's (1997), Andersen's (1997) and
Liu's (1994), which give a framework and techniques whereby the principles
of semiotics can be applied to information systems design.
2.1 Information, data and meaning
Our approach to this problem is to look at the origin of the problem,

namely the relationship among information, data and meaning with a
semiotic perspective.
Mingers (1995) identifies the objectivist view and the subjectivist view of
information. The former takes information to be 'processed data', and the
latter takes the view that information equals data plus meaning attributed to
them by a receiver of the data (Checkland 1981, Checkland and Holwell
1998, Stowell and West 1994, Lewis 1994). Both views have weaknesses
(Mingers 1995). To develop a systematic treatment of this problem,
Mingers' (1995) framework of sign, information and meaning, Drestske's
(1981) semantic theory of information seem relevant and insightful and fit
with semiotics (Stamper 1997, Andersen 1997) well. Dretske's theory gives
us the origin, quantity, and content of information, the concepts of
A Semiotics Approach to Analyzing the Information Bearing... 235
'informational relationship', and' information content of a signa/'o Mingers'

(1995) framework clearly defines the interrelations among sign (including
data), information and meaning. Mingers defines three levels of meaning.
Semiotics looks at the world in terms of signs and their properties. The
syntactics and semantics levels of the properties are particularly relevant to
the problem at hand. We take these as the foundations of this work, bases
upon which, a set of basic concepts are developed, which are summarised
below.
Following Dretske (1981), information will be taken as created by or
associated with a state of affairs among a set of possibilities of a general
situation, the occurrence or realization of which reduces the uncertainty of
the general situation. The quality of the information that is created by or
associated with a state of affairs is the actual reduction in uncertainty, which
can be measured by using probability theory. Information is therefore seen
as an objective commodity. The content of information may be seen as a
state of affairs (a situation) and what is or will be true in that state of affairs.
Information can be transmitted. A state of affairs, say r" is a particular
case or an instantiation of a general situation, say r. The reduction in
uncertainty at r due to the occurrence of r, may be accounted for by one or
more events, say s" S), ... ,S", that occur at another general situation, say S.
This gives rise to a special kind of relationship - 'informational relationship'
(Dretske, 1981, p.3 5) - between these two general situations rand S. An
informational relationship captures certain degree of dependency between a
state of affairs r, of a general situation r and what takes place in another
general situation S. This dependency can be demonstrated by the fact that
r/s appearance alters the distribution of probabilities of the various
possibilities at S. The dependency is a type of regularity concerning different
general situations based upon nomic dependencies (Dretske, 1981), logic, or
norms, etc. in a social setting. Due to this relationship, information created at
s is transmitted to r. We will call s the 'information source', and r 'the
receiver of information about s'. Moreover, a state of affairs r, at r can be
seen as a signal that carries information about S. A sign carries information
about states of affairs in the world - what it signifies, even though the sign
may never be actually observed by anyone. Besides, if it is recorded, r,
becomes a piece of data. Thus data carry information. In general, data in an
information system are a collection of recorded signals or events at the
receiving end of the information transmission, which provide information
about the source. Information is carried by a sign and is objective and in
analogue form.
Information is literately untouchable as one is always already in a system
of meaning (Mingers 1995). Meaning is the semantic content of a sign
generated from the available information and obtained by an agent, 'not just
236 Junkang Feng
the meaning of a word or sentence', but 'meaning quite generally' (Barwise

and Perry 1983), and meaning is inter-subjective or subjective and in digital
form. Here 'digital' means being discrete, precise and well bound. It would
seem reasonable and certainly useful to take meaning as made up of items.
And each item of meaning may be seen as regarding a state of affairs, which
in turn can be seen as a relation among a number of objects.
The above basic concepts provide us with a means to tackle this problem
with theoretic rigor. When applying these ideas to our problem, it would
seem neat and beneficial to adopt a semiotic perspective (Stamper 1997,
Andersen 1997). Stampers' organisational semiotics enables us to take 'sign'
as a primitive concept in pursuing a truly scientific study of information
systems. Properties of a sign can be divided and organized into six layers
from physical means, empirics, ... , to social world. Of the six layers, the
syntatics and semantics layers and the dependencies between them would
seem particularly relevant to this study. According to the ideas of semiotics,
an information system can be viewed as a sign system. And the information
that is represented by the system is some properties of the signs in the
system. So it would seem necessary to look at what determines the structure
of the signs in the system. The signs in such a system are normally called
data, and a data schema specifies the structure of data. Therefore the schema
ultimately determines the capacity of the information system's of providing
information. Based upon this reasoning, we would like to propose the notion
of 'information bearing capacity' of a conceptual data schema as a vehicle
for tackling the problem at hand. Furthermore it would seem plausible to
define this capacity to be the collection of items of meaning that can be
obtained by agents from the data specified by the data schema. Items of
meaning converted from information may be unlimited due to 'information
nesting' (Dretske 1981, p. 71). From the point of view of information
systems design however, we want to make sure that the items of meaning
that can be obtained from stored data include those that the system is
designed to provide.
To identify its 'information bearing capacity', the characteristics of a
conceptual data schema that are relevant to this capacity have to be
examined. To this end, Stamper's (1997) observation is insightful, i.e., 'signs
on every level depend for the correct formation of signs on the level below'.
A data schema should therefore be analyzed at different levels, among which
the 'syntactic' and 'semantic' seem the most relevant. The former is
concerned with the formal structure of a schema, and the latter objects and
relationships among them that the signs (i.e., data) and structures of the signs
signify. So the analysis of the 'information bearing capacity' becomes a
matter of examining the relationship between the two levels. In the rest of
the chapter, we will use data schemas in ER to illustrate our points.
A Semiotics Approach to Analyzing the Information Bearing 237
Capacity of a Data Schema
matter of examining the relationship between the two levels. In the rest of
the chapter, we will use data schemas in ER to illustrate our points.
2.2 Topological connections vs. semantic relations
As we said in the last section, of the six layers of properties of signs, the
syntactics and semantics, would seem more relevant to our problem. And we
aim to reveal the dependencies between these two layers in the context of an
ER schema.
It may be seen that there are two types of connections between entity
instances. The connections that are made possible by the topological
structure (i.e., a syntactic level formation of signs) of an ER schema can be
termed 'topological connections' regardless of their meaning. The reason
why we suggest to call them 'topological connections' is because at the
syntactic level, the connections of data constructs are of a topological nature.
That is to say, the connections remain unchanged even when a diagram for
the ER schema is changed (deformed) by bending, stretching and twisting
etc. The connections in terms of meaning can be called 'semantic relations',
which is the meaning (i.e., digitalised information) we want to represent by
using some 'topological connections'. 'Semantic relations' are independent
of a modeling mechanism such as ER. For example, suppose that 'employee
e I belongs to division d l' is a true proposition, we say that there is a
semantic relation 'belongs to' between two entity instances employee el and
division dl. If these two entity instances are topologically connected in an
ER schema, we say that there is a topological connection between them. A
basic task in information design is to construct a sufficient (minimally
sufficient if possible) data schema that enables all semantic relations that are
required to be represented to be actually represented by topological
connections that are made possible by the schema. In order to achieve this,
we must understand what is meant by a topological connection representing
a semantic relation. This takes a few more related concepts to define.
2.2.1 A topological connection corresponds to a semantic relation
A topological connection, say t, corresponds to a semantic relation, say s,

if both are concerned with the same entity instances, the same instances are
connected in both, and the meaning of t can be interpreted as that of s. For
the example in Figure I, student s2 and subject c I are connected, which
forms a topological connection. If it can be interpreted as 'reads', then this
topological connection corresponds to the semantic relation 'student s2 reads
subject cl'.
238 Junkang Feng
2.2.2 Irrelevant and relevant topological connections
Student Subject
Student
Subject
sl
c1
s2o---~-------+~~
s3O----r______~~c2~}
s4
Figure 1. Irrelevant and relevant topological connections
Against a set (a type) of semantic relations, a topological connection is an

irrelevant topological connection if it does not correspond to any semantic
relation of this type. For example, 'an undergraduate student reads a subject'
is a set of semantic relations. Assume that in the topological connection (s 1,
c 1) in Figure 1 s 1 is a postgraduate student, then this connection does not
correspond to any semantic relation of 'an undergraduate student reads a
subject', so it is an irrelevant topological connection against the given set of
semantic relations. On the contrary, if it corresponds to a semantic relation, a
topological connection is a relevant topological connection with regard to
the semantic relation.
2.2.3 Distinguishable topological connections
Let schemal be an ER data schema, t a topological connection made

possible by schemal, S a set of semantic relations and s an instance of S; and
let t correspond to s. t is distinguishable regarding S if S can be explicitly
defined by using the components of the data schema schemal. Note that if
all irrelevant topological connections can be explicitly defined by using the
components of the data schema schemal, S can also be explicitly defined as
a consequence.
lftTkS for 1_ _--...

(1,1) (n,m
Lecturer Department Faculty
Figure 2. Relevant topological connections that can be explicitly defined
For example, for the path in Figure 2, assume that only full time lecturers
belong to a faculty, and they belong to the faculty under which the
department they work for is. With regard to the semantic relation 'a lecturer
belongs to a faculty', all topological connections involving a part time
lecturer and a faculty that are made possible by the path are irrelevant ones.
Of all the possible topological connections, as long as those that correspond
to 'a full time lecturer belongs to a faculty' can be defined by, say, the post
of a lecturer, the hours per week they work, etc, then the relevant topological
connections are distinguishable. That is, a full time lecturer might be defined
as:
Full time lecturer = Lpost ~ FT Lecturer, or
Full time lecturer works for Department = Lhours > 35 (Lecturer works for
Department)
2.2.4 A topological connection represents a semantic relation
Let schema] be an ER data schema, t a topological connection made

possible by schema], S a semantic relation type, and s an instance of S. t
represents s if t corresponds to t and t is distinguishable regarding S.
For example, suppose that in Figure 3, topological connection (el, dpl,
d 1) is relevant while topological connection (e2, dp 1, d 1) is irrelevant.
Assume that el and e2 do not belong to different proper subsets of the entity,
then neither (el, dpl, dl) nor (e2, dpl, dl) can be explicitly defined.
Consequently the relevant topological connection (e 1, dp 1, d 1) cannot be
differentiated from the irrelevant topological connection (e2, dpl, dl).
It should be noted that if there is no irrelevant topological connection in a
path with regard to a type (set) of semantic relations, then the question of
whether a topological connection is distinguishable does not arise. That is,
all topological connections represent that set of semantic relations. We will
say more about this later.
240 Junkang F eng
Employee Department Division
-
elo-
dpl dl
e2
e30
e40
Figure 3. A topological connection does not represent a semantic relation due to being non-
distinguishable
The above discussion also shows that a 'representing' topological

connection must be a 'corresponding' one first. But the reverse is not true.
Figure 4 illustrates this point, where t is a topological connection made
possible by a data schema, S a semantic relation type, s an instance of S.
t vs. s
t corresponds to s (t is relevant to S)
t represents s
)
t does not corresponds to s (t is irrelevant to S)
Figure 4. A topological connection corresponds to or represents a semantic relation
There are certain types of semantic relation(s) that the topological

connections of a path can always unconditionally represent (also corresponds
to, by definition), we will call these semantic relations 'primary meaning' of
the path.
3. TYPES OF MEANING OF A PATH
3.1 Primary meaning
For a path in an ER schema, a topological connection made possible by

the path always has a primary meaning. For example, the path in Figure 5
has the primary meaning that a lecturer delivers a lecture, and a student
attends a lecture. These are the semantic relations that the topological
connections can always correspond to and represent with no conditions.
Lecturer Student
Figure 5. Primary meaning vs. implied meaning of a path
3.2 Implied meaning
Given certain conditions on both the topological level and the semantic
level, a topological connection may represent a semantic relation that is
beyond the primary meaning. For example, the path in Figure 5 is capable of
representing 'a lecturer lectures a student', in addition to the primary
meaning that we have just said.
3.3 Conditions for the creation of implied meaning
For the conditions on the semantic level, we look at organizational rules

and the logic of a matter. If a lecturer delivers a lecture, and a student attends
the lecture, then the lecturer lectures the student. This is logical. In an
organization, there might be an organizational rule, namely 'an employee
may only work on a project that is controlled by the department the
employee belongs to'. Then from 'an employee works on a project' and 'a
project is controlled by one (only one) department', we get 'an employee
belongs to a department' .
For the conditions on the topological level, we look out for the structure
of a path. Due to its particular structure, a path might not be able to provide
topological connections that correspond to a given set of semantic relations,
or a path is capable of providing corresponding topological connections but
they are not distinguishable. We pay attention to the length of the path, the
participation constraints of the entities, and so on. When the length of a path
is greater than one, we watch out for those situations where the 'plurality of
242 Junkang Feng
joins' (Codd 1970) may apply. Codd's concept of 'plurality of joins' neatly
captures the type of situations where two or more relations are involved or a
path of length > 1 is used and there may be irrelevant topological
connection(s). Moreover these irrelevant topological connections provide a
condition for a relevant topological connection being not distinguishable
from them. This prevents the relevant topological connections from being
able to represent the semantic relations that they correspond to.
To systematically examine whether an item of meaning can be
represented and therefore obtainable for an agent, classifying a path with
regard to an item of meaning, i.e., a set of semantic relations seems useful.
4. CLASSIFICATION OF A PATH
With regard to a given set of semantic relations between entities within it,
by virtue of set theory and the ideas about connection traps (Codd 1970,
Howe 1989), a path can be classified. We will describe them one by one. In
the discussion, we will use set A to represent a set of topological connections
between some entities that are made possible by the structure of a path, and
set B to represent a set of semantic relations between these entities. A always
represents some semantic relations (see earlier on what is meant by 'a
semantic relation is represented by a topological connection'), but not
necessarily B. In the discussion below, we will use set operations. The
meaning of the operations, such as 'n' and 'c' are based upon 'corresponds
to' not 'represents'. (These two concepts were defined earlier). For example,
B c A means that all members of B are corresponded to by members of A,
and at least one member of A does not correspond to any member of B.
4.1 Classify A regarding B in terms of whether A

corresponds to B
We will first classify a path in terms of whether it can have topological

connections to correspond to a given set of semantic relation. On this level
when corresponding topological connections can be available we will
assume that they are non-distinguishable from irrelevant ones. In what
follows, the numbering of the classes is arbitrarily chosen (i.e., Class A,
Class B 1, ... ) except that Bland B2, and C 1 and C2 have some common
features. The names of the classes, however, do reflect the capacity of a
path's in corresponding to and/or representing a semantic relation.
Class A (mutual exclusion)
DO Figure 6. Class A
Figure 6 visualizes the essence of a Class A path, namely An B = 0. No

member of B is corresponded to by any member of A. So no member of B is
represented by that of A. For example, it seems reasonable to assume that
path(supplies, uses) in Figure 8 captures no information on 'which supplier
is geographically near the headquarters of which projects'. For this example,
A = 'A supplier supplies a part, and a project uses a part',
B = 'A supplier is geographically near the headquarters of a project' .
So with regard to this information expressed as B, the path is of 'Class A
(mutual exclusion),.
!70.Plies 1_ _---...
(1,n) (1,1)
Supplier Part Project
Figure 7. A path of 'Class A (mutual exclusion), with regard to 'which supplier is

geographically near the headquarters of which projects'
Class B 1 (partial and non-distinguishable inclusion)
A :c-:
I
I
I
I
B
I I
I I
.... --~
Figure 8. Class B 1
Figure 8 illustrates what determines that a path is of Class B 1. The figure

shows that A n B = C =1= 0, A tt B, and B tt A. Members of D do not
correspond to any members of B. That is, D are irrelevant topological
connections, which may give rise to 'false semantic relations'. D may be
caused by a 'fan trap' (Howe 1989). Members of D and members of C
cannot be differentiated. E are members of B that have no corresponding
topological connection of A. This may be caused by a 'chasm trap' (Howe
1989). We will show an example below.
244 Junkang F eng
Lecturer Department Course
Figure 9. A path of 'Class B I (partial and non-distinguishable inclusion), with regard to

'which lecturer is involved in the teaching of which courses'
With regard to 'which lecturer is involved in the teaching of which

courses run by a department', path(belongs-to, runs) in Figure 9 does not
capture, say, part-time lecturers' involvement in the teaching of the courses
run by a department, as they do not belong to the department in that they are
not formally on the staff of a department. The path cannot capture 'which
full-time lecturer is involved in the teaching of which courses run by the
department' either, as there is a fan trap in the path. But the path does have
some topological connections that correspond to the correct information, but
one just cannot tell which ones. In this example,
A = 'A lecturer belongs to a department, and a department runs a course',
B = 'A lecturer is involved in the teaching of a particular course run by a
particular department' ,
C = 'A lecturer who belongs to a department is involved in the teaching
of a particular course run by this department',
D = 'A lecturer who belongs to a department is not involved in the
teaching of a particular course run by this department',
E = 'A lecturer who does not belong to a department is involved in the
teaching of a particular course run by a department' .
A is the union of C and D. B is the union of C and E. C and D cannot be
differentiated. So with regard to this information expressed as B, the path is
of 'Class B I (partial and non-distinguishable inclusion),.
Class C I (total and non-distinguishable inclusion)
A ,--, ,- - ....
,,,0',, IB
,, ,, I
.... --~ \
Figure JO. Class C I
In Figure 10, B i A. Members of D do not correspond to members of B.

That is, D are irrelevant topological connections, which may give rise to
'false semantic relations'. D may be caused by a fan trap. Any member of B
A Semiotics Approach to Analyzing the Information Bearing. .. 245
has a corresponding member of A, there is no chasm trap. But members ofD

and members ofB cannot be differentiated.
For the current classification, only those situations where A mayor may
not correspond to B, and if A does, no corresponding topological connection
is distinguishable from irrelevant ones are considered. We would like to
draw the attention of the reader to the fact that apart from the above three
classes there can be no any other possible classes for a path under our
present premise. This is because if no irrelevant topological connection
exists in a path with regard to a set of semantic relation, then a
corresponding topological connection cannot be non-distinguishable from
irrelevant ones.
d) (lJ ~) e)~
F"I
Figure 11. Classes where a path cannot have irrelevant topological connections
The only other possible relationships between A and B are the two
illustrated in Figure 11. Both d) and e) in it shows that all topological
connections of A correspond to semantic relations of B, so there is no
irrelevant topological connection in A regarding B. So it is not possible that
in these two situations A correspond to B but relevant topological
connections of A are non-distinguishable from irrelevant ones.
4.2 Classify A regarding B in terms of whether A

represents B
Now we will look at all possible relationships between A and B in terms

of whether A represents B. First of all, the three classes introduced in the
previous section, namely classes A, Bland Clare the same in that if it is of
any of these three classes regarding a set of semantic relations then a path is
not capable of representing this set of semantic relations. Now we will
define those classes that are capable of at least partially representing a set of
semantic relations.
Class B2 (partial and distinguishable inclusion)
246 Junkang Feng
Figure J2. Class B2
Figure 12 shows that A (\ B = C "* 0, Act. B, and B ct. A. Members of F

do not correspond to members of B. That is, F are irrelevant topological
connections in A. Members of F and members of C can be differentiated. E
are members of B that do not have corresponding topological connections in
A, which may be caused by a chasm trap.
Projects
[ Software]
[Hardware]
CPl
(0, n)
CPl
Staff Department
[ Fname ]
Other
Figure J3. A path of 'Class B2 (partial and distinguishable inclusion), with regard to 'which
engineer works for which department
We are now giving an example by using Figure 13. Note that the figure is
a higraph (Harel 1983) version of an ER schema. The dotted lines in CP1
indicate that CP1 is the Cartesian product of NIN, Sname, etc. And entity
Staff is a subset of CPl. This is a way to capture attributes of an entity in
terms of higraph.
For the path in this figure, initially A is 'a member of staff works on a
project, and a department controls a project'. If there is a rule, namely 'if a
member of staff works on a project, and the project is controlled by a
department, then the member of staff works for the department', then
path(works on, controls) in this figure represents 'a member of staff who
works on a project works for the department that controls the project'.
Now let us take B to be 'an engineer works for a department', then
C = 'An engineer who works on a project works for the department that
controls the project' ,
A Semiotics Approach to Analyzing the Information Bearing. .. 247
E = 'An engineer who does not work on a project works for a

department' ,
F = 'A member of staff other than engineer who works on a project
works for the department that controls the project' .
B is the union of C and E. A is the union of C and F. C and F can be
differentiated by using attribute Post of entity Staff. So the path in Figure 13
represents part of B. That is, the path is of Class B2 (partial and
distinguishable inclusion) with respect to 'which engineer works for which
department' .
For the rest of the classes, due to length constraints no examples will be
given.
Class C2 (total and distinguishable inclusion)
Figure 14. Class C2
Figure 14 illustrates that B c A. Members of F do not correspond to

members of B. F are irrelevant topological connections. Members of F and
members of B can be differentiated. Any member of B can be corresponded
to by member(s) of A. There is no chasm trap.
Class D (partial representation)
Figure 15. Class D
Figure 15 shows that A c B. Any member of A corresponds to a member

of B, so there is no irrelevant topological connection in A. Therefore all A
represent B. E are members of B that have no corresponding topological
connection from A. That is, E are those semantic relations that cannot be
represented by A. E may be caused by a chasm trap.
Class E (total representation)
o
248 Junkang Feng
Figure J6. Class E
Figure 16 shows that A = B. There is no irrelevant topological connection

in A with respect to B. There is no fan trap or any other false semantic
relation in A. So any member of A represents a member of B. On the other
hand, there is no chasm trap, and in general any member of B has its own
distinguishable corresponding topological connection{s) from A. That is, all
members of B are represented by A.
From the definitions and examples of the classes of paths shown above,
we can see that in order to represent a set of semantic relations between
entities, with respect to this set of semantic relations, one of the following
must be found:
a path of 'Class C2 (total and distinguishable inclusion)'
a path of 'Class E (total representation)'
- a group of paths of 'Class B2 (partial and distinguishable inclusion), or
'Class D (partial representation)" the union of which represent the whole
set of semantic relations.
A deduction can therefore be reached, i.e., if it is of a class other than B2
(partial and distinguishable inclusion), C2 (total and distinguishable
inclusion), D (partial representation) and E (total representation) with respect
to a set of semantic relations, a path is of no use in terms of representing the
set of semantic relations.
5. THE TYPES OF MEANING OF A PATH VS. THE

CLASSES OF A PATH
Now we are in the position to show the relationships among the three
terms used extensively in this chapter, namely, 'organizational rule and the
logic of a matter', 'meaning of a path', and 'class of a path'. Figure 17
illustrates them by using a higraph.
Classes of a path
I
I Useless classes
I
I
I
Organisational
rules orland logic
of the matter, and -----
the structure "', Useful classes
.~
0 @]
ofa path
._.-.~.. -.-. .[D @J
Basic meaning
Figure J 7. The types of meaning of a path vs. the classes of a path
The 'basic meaning' of a path is the set of semantic relations with respect
to which the path is of Class E (total representation) without referring to any
organizational rules or the logic of the matter that the path is concerned with.
Given the particular structure of a path, applying 'organizational rules orland
the logic of the matter' on the basic meaning of a path may result in some
implied meaning of the path. An implied meaning of a path is either part or
the whole of a set of semantic relations that the path is capable of
representing. That is, a set of semantic relations with respect to which the
path is of one of the useful classes, namely Classes B2 (partial and
distinguishable inclusion), C2 (total and distinguishable inclusion), D
(partial representation) or E (total representation). Besides, a path can never
be a connection trap as far as using its 'basic meaning' goes, provided that
the basic meaning is unambiguously defined. Note that an item of implied
meaning may give rise to another item of implied meaning when some
rule(s) orland dependencies etc. are applied to the former. This can happen
because given the former, the latter's conditional probability is 1, otherwise
less than 1. This idea is drawn from Dretske's (1981) definition of
'informational content of a signal', and the idea of 'information nesting'.
This point is illustrated in Figure 17 with the recursive relationship of the
'implied meaning'.
Having all the above in place, the' information bearing capacity of a data
schema' can be measured as the collection of the semantic relations for each
of which there is at least one path being of a useful class in the schema. Note
that we have omitted those trivial cases where an item of meaning is
represented by an entity or an attribute.
250 Junkang Feng
6. USING THIS METHOD IN TACKLING FAN

TRAPS
In this section we will use an example to show that the concepts and
method described in this chapter can be used to tackle fan traps.
A type of fan traps is where the relationship affected by the fan trap
involves one entity, say A, and a subset, say B 1, of another entity, say B, and
no more. Such a path will inevitably include a recursive relationship. Figure
18 shows an example. Assume that a team leader is responsible for the
projects that hislher team members work on, then path(leads, works-on)
might appear to capture a relationship between Team Leader, Team Member
and Project. But if you use the path to capture 'who is responsible for which
projects', you will fall into a fan trap as one engineer may work on different
projects within different teams. The diagram at the instance level in Figure
18 shows the situation.
t;5.rks-on 1_ _--..
(O,n) (l,n)
Engineer Project
Team Team
leader member
Engineer
Project
j1
j2
tl2
Figure 18. A fan trap involving a recursive relationship
To avoid the fan trap, we can re-structure the ER schema to be the one
illustrated in Figure 19.
!ii.rks-on ,_ _--...
(O,n) (I,n)
Team Member Project
(I,n)
(I,n)
Team Leader
Figure J9. Removing a recursive relationship
The point of so doing is to represent the subset affected by the fan trap
explicitly. As a result, path(leads, works-on) in Figure 19 clearly shows a fan
trap. To represent 'who is responsible for which projects', we need a sound
and complete path for it. We can add a non-transitive relationship
'responsible-for' between Team Leader and Project to solve the problem as
this path is a Class E (total representation) path. This is shown in Figure 20.
Incidentally this 'responsible-for' relationship in fact also helps the
formation of a Class E path, namely path(works-on, responsible-for), for the
information of 'who works on which projects that is responsible for by
whom'.
!ii.rks-on ,_ _--...
(O,n) (l,n)
Team Member Project
(I,n)
(I, I ) .....J responsible-for I
(I,n)
(I,n)
Team Leader
Figure 20. Adding a sound and complete path to avoid a fan trap
252 Junkang Feng
7. SUMMARY
To be able to identify the capacity of a formal information system for
providing information is of great importance and highly desirable, and yet it
does not seem to have been well defined or addressed in the literature. In this
chapter the notion of 'information bearing capacity' of a conceptual data
schema has been proposed for tackling this problem. This work is based
upon known theories on the semantic aspect of information, and its relations
with meaning and data, which were used from a semiotic perspective. The
approach adopted in this chapter enables the separation of topological
connections that are made possible by a path and the semantic relations that
are required to be represented by the path. This chapter shows that this
approach is insightful and effective, and has enabled us to achieve our
objectives.
The main results, in addition to the approach per se, of the work reported
in this chapter are the following:
A set of concepts for our approach: a topological connection corresponds
to a semantic relation; a topological connection represents a semantic
relation; irrelevant and relevant topological connection with respect to a
set of semantic relations; distinguishable relevant topological connection;
the primary meaning and implied meaning of a path.
The concept of 'classes' of a path. We have shown that there are only
three classes that a path can have in terms of whether a path is capable of
providing topological connections to correspond to but not represent a
set of semantic relations. In terms of whether a path is capable of
providing topological connections to represent a set of semantic relations
there can be five classes. Of these classes, only four of them are useful.
To examine the 'information bearing capacity' of an ER schema is to
identify all semantic relations for each of which there is at least a path
being of a useful class in the schema.
REFERENCES
Andersen, P. B. 1997, A Theory of Computer Semiotics, Cambridge University Press,
Cambridge.
Avison, D.E., and Shah, H. U. 1997, The Information Systems Development Life Cycle: A
First Course in Information Systems. London, McGraw-Hill.
Batini, c., Ceri, S., and Navathe, S.B. Conceptual Database Design: An Entity-relationship
Approach Redwood City in Calif, Benjamin/Cummings, 1992.
Barwise, J., and Perry, J. Situations and Attitudes. Cambridge Massa, Bradford Books, MIT
Press, 1983.
Checkland, P. 1981, Systems Thinking, Systems Practice, John Wiley & Sons, Chichester.
Checkland, P., and S. Holwell, 1998, Information, Systems and Information Systems, John
Wiley & Sons, Chichester, New York
Codd, E.F., 1970, A Relational Model of Data for Large Shared Data Banks. Com. of ACM,
13.6, pp.377-387.
Dretske, F. I. Knowledge and the Flow ofInformation, Oxford, Basil Blackwell, 1981.
Eick, C. F and Lockemann, P. C., 1985, Acquisition of terminological knowledge using
database design techniques, Proc ACM SIGMOD 1985 International Conf on Management
of data, SIGMOD Record Vol 4, No4 p.84.
Flynn, D., 1998, Information Systems Requirements: Determination & Analysis 2nd ed.
London, McGraw-HilI.
Harel, D., 1988, On Visual Formalisms, Com. of ACM, 31.5, pp.514-30.
Howe, D.R., 1989, Data Analysis for Data Base Design, 2nd ed, Edward Arnold, London.
Lewis, P., 1994, Information-systems Development, Pitman, London.
Liu, K., Yasser, A. and Stamper, R., 1994, Simplicity, uniformity and quality - the role of
semantic analysis in systems development, in Managing Quality Systems, Ross et al. Eds.,
Computational Mechanics Publications, Southampton.
Mingers, J., 1995, Information and meaning: foundations for an intersubjective account, Info
Systems J, 5, 285-306.
Moody, D., 1998, Metrics for evaluating the quality of entity relationship models,
Proceedings of 17th International Conference on Conceptual Modeling (ER'89),
Singapore, 16-19 November.
Stamper, R., 1997, Organisational semiotics, in: J. Mingers and F. Stowell, ed., Information
Systems: An Emerging Discipline? McGraw Hill, London, pp.267-83.
Stowell, F. and West, D., 1994, Client-led Design, McGraw-Hili, London
Wilson, B. Systems: Concepts, Methodologies and Applications 2nd ed. Chichester, Wiley,
1990.
Index
actability, 88, 95, 10 1, 102, 104, 105, denotation, 173

109,110,111,112,113,114,115,131 deontic operators, 228, 232
Actants, 135 digitalised information, 243
action context, 96, 112, 121, 122, 123, discourse, 7, 9,17,18,67,100,101,138,
126 153,165,172
action layers, 8 electronic commerce, 117, 130, 172
actuality, 162, 164 exterior facet, 12, 13, 14, 15, 16, 17, 18
basic meaning, 256 Functional Grammar, 13,21,86
business actions, 105, 125, 130 functionalist, 160
Castor-project, 167 genre, 20, 63, 67, 68, 70, 71, 72, 75, 76,
cognition, 116, 153, 157, 161, 188,214 77,78,79,82,84,165
Communication acts, 122, 221 Greimas' narrative semiotics, 134
communication channel, 63 historical, 64, 160, 162
communication quality, vi, 117, 119, 120, human communication, v, 3, 72, 87, 89,
126, 127, 129, 130 114, 119, 218, 220, 221
communication types, 36 human resources, 171
communicative action, 8, 89, 93, 95, 98, illocution, 88, 90, 91, 95, 100, 114, 120
99,105,110,115,118,119 implied meaning, 247, 248, 256, 259
communicative functions, 32, 49, 50, 57 information bearing capacity, vi, 239,
communicator, 103, 111 242,256,258,259
computational, 160,235 information content, 128, 240, 241
conceptual strategy, 133 information systems actability, 88, 89,
concrete manifestation, 6, 7 114
coordination, v, vi, 23, 24, 25, 26, 29, 30, infra layers, 6, 9
31,33,37,43,45,46,47,48,49,50, instrument, 42, 51, 92, 93,168,198,199,
51,52,53,54,55,56,57,58 200
core layers, 6, 8, 9 intention, 25, 95, 96, 100, 150, 152, 184,
CSCVV,23, 25, 51, 58, 59,237 222,226,230,233,235
data schema, 239, 242, 243, 245, 246, interdependence, 25, 29, 30
256,258 interdependent actors, 25
256 Index
interior facet, 12, 13, 14, 15, 16, 17, 18, phenomena, 6, 10, 11, 12, 15, 17, 18,25,
19 97,157,164,186,197,226
interpreter, 90, 103, 111, 122, 126, 181 phenomenological, 133, 161, 187, 188,
interpretive, 133, 153, 160 215
intersubjective, 165,260 phonological analysis, 11
knowledge change, 167 phonology, 15
knowledge conversion, 167, 168 physical resources, 171
language action, 87, 90, 10 1, 115 pragmatic factors, 7
maritime operations, 24, 26, 29, 31,55 pragmatic perspective, 88, 114
memory, 78, 102, 162, 175, 180 primary meaning, 247, 248, 259
modality, 24, 32, 34, 35, 36, 37, 38, 49, proposition, vi, 39, 88, 230, 236, 243
50,71,144,147,150,152 protocols, 53, 54, 111, 180
mode, 63,100,101,105,108,110,114, Register, 63
144, 145, 146, 152 representational strategy, 158, 161
modulation, 71 role schema, 234, 235
multi-actor system, 157, 169 sales support systems, 126
multi-agent system, 218, 220, 228, 235, semantic model, 219, 222, 223, 226, 227,
237 228,230,231,235,236
narrative analysis, 134, 138, 151 semantic relations, 240, 243, 244, 245,
narrative program, 134, 135, 136, 137, 246,247,248,249,250,251,252,
138,142,143,144,146,151 254,255,256,258,259
Natura11anguage, 3 semiotic dimension, 158
navigation, 27, 51 sensemaking, 138, 165, 187, 191, 193,
norms, 19,20,22,25,39,40,41,94,111, 196, 197, 215
116,120,128,130,176,219,222, signification, 6, 173, 187, 188, 190, 196,
223,225,226,227,228,230,231, 211
232,233,234,235,236,241 social interaction, 121, 123
behavioral norms, 39 Social Semiotics, 58, 79, 82, 86
behavioural norms, 226 societal resources, 171, 177, 178
cognitive norms, 226 software agents, 217, 218, 219, 220, 223
intrinsic norms, 39, 232, 234 speech act theory, 87, 90, 91, 94, 95,114,
perceptual norms, 225 118,119,120,128
valuative norms, 226 stakeholders, 65, 79,191,192,194,199,
obligations, 47,71,108,118, 127,228 203,205,211,212
ontology, 135, 152, 153, 221 stress, 10, 15, 74, 104, 117, 185, 186,
Operation Dialogue, 139, 141, 142, 143, 187,188,189,190,191,192,195,
144,145,146,147,148,149,151 200,205,206,207,208,209,210,
organisational behaviour, 158, 164, 217 212,214,215
organisational representation, 159, 163 syntactic description, 10
organisational strain, 192, 193, 195, 196, systemic functional linguistics, 61, 64, 76
205,212 systemic semiotics, 61, 64
organisational structures, vi, 133, 150, tenor, 63, 82
152 topological connections, 239, 243, 244,
organising grammar, 142 245,246,247,248,249,250,251,
paradigm, 34, 37, 67, 153,220 252,254,258,259
performer, 103, 105, 110, 111, 126 ultra layers, 6, 9
perlocution, 88, 90, 95, 120 utterance, 6, 32, 34, 35, 36, 38, 40, 79,
Person-environment fit, 186, 214 88,89,92,95,97
person-environment interactions, 188

Coordination and Communication Using Signs:: Studies in Organisationat Semiotics 2

Uploaded by

Copyright:

Available Formats

You might also like

Coordination and Communication Using Signs:: Studies in Organisationat Semiotics 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Coordination and Communication Using Signs:: Studies in Organisationat Semiotics 2

Uploaded by

Copyright:

Available Formats

Coordination and Communication

INFORMATION, ORGANISATION AND TECHNOLOGY: Studies in Organi-

Springer Science+Business Media, LLC

July 2000, hosted by Staffordshire University"--Pref.

HD30.2 .1558 2002

Copyright © 2 0 0 2 Springer Science+Business M e d i a N e w Y o r k

Permission for books published in Europe: permissions@wkap.nl

Printed on acid-free paper.

PART 1: COMMUNICATION AND PRAGMATICS

1. Accomodating Natural Language Within The Organisational Semiotic

3. Intertextuality at Work: Large Scale Organisation of Workpractices 59

4. Actability: a Way to Understand Information Systems Pragmatics 85

5. Communication Quality in the Context oflnformation Systems and

6. Greimas' Semiotics and the Analysis of Organisational Action 129

PART II: ORGANISATIONAL SYSTEMS 151

7. Reconsidering the Standard: a Semiotic Model of Organisations 153

8. Using Semiotic Framework to Study Social Consequences ofIT 167

9. Implementing IT in Production Settings 179

10. A Semiotic Approach to Improve the Design Quality of Agent-Based

11. A Semiotics Approach to Analyzing the Information Bearing Capacity

ANDERSEN, PETER B0GH

Organisational Semiotics occupies an important niche in the research

, Details of the mini series can be found at http://www.cs.rdg.ac.uk/~sisOlkl

the context of organisations, whether or not they use information technology.

Owen Eriksson, Michael Heng, Joaquim Filipe, Vasser Ades, Samuel

Kecheng Liu, Rodney Clarke, Peter B. Andersen and Ronald Stamper

The Programme Committee

Albert Alderson, Staffordshire University, United Kingdom

COMMUNICATION AND PRAGMATICS

Accomodating Natural Language Within The

Natural language (NL), by virtue of its central role in human

electronic or paper means, the production of reports on past, present and

automatic translation programs which might be employed to help cope with

Figure 1. Stamper's Organisational Semiotics framework

2.1 The Existing Framework

The OS framework recognises six levels or layers, as shown in Fig 1.

(b) Pragmatics: speaker is opening a new section of dialogue, posing

2.2 Alternative Groupings

Figure 2. An enhanced version of the Organisational Semiotic framework

This proposal is reflected in Fig 2, which is a modified version of Fig 1.

communicated, their interpretation forms part of the pragmatics of the

2.3 Benefit of the Alternative Groupings

of these important factors would escape the designer's net. Furthermore,

3. SUBDIVIDING THE LAYERS

3.1 Accepted Subdivisions

In order for the OS framework to reflect the structure of NL in more

To treat 'come across' here as merely a combination of verb plus

3.2 Subdividing the Layer of Empirics

'Empirics' is not a term that is traditionally used in the description ofNL.

3.3 The Interior and Exterior Facets of the Layer of

The distinction between interior and exterior facets turns out to be

the level of syntactics can be described, at least in part, in tenns which do

might be taken as either a semantic or a pragmatic distinction. In either case,

3.4 The Interior and Exterior Facets of the Layers of

Semantics, like syntactics, includes certain phenomena which are

3.5 Benefit of Recognising Layers and Facets

NL-based systems can be complex, and so a major advantage of dividing

4. PARALLELS BETWEEN LAYERS