Tim French and Andy Smith

University of Luton, Park Square, Luton LU1 3JU, UK

The design of any web-site is critical to the success of the site and to the meanings, both intended and unintended it may be transmitting. We believe web-sites to be particularly sensitive to issues of trust and security as well as to cross-cultural issues. How are we to design sites which transmit only intended meanings to users and across cultures? We believe that traditional methods of analysis and interface design are no longer fully appropriate. Rather, the design of web-sites are more like creating a film or other mass media for an ill-defined audience. Thus tools used for analysing mass media would be more appropriate here: one such tool is semiotics.

Semiotics has been called the ‘science of signs’. More specifically it is the discipline which connects meaning, meaning making, communication and culture through an understanding of acts of signification. There are a variety of semiotic discourses and traditions, most noticeably that of Peirce (1953) and Saussure (1974) and there have been various attempts to apply a semiotic framework in the field of IT and HCI (French, Polovina and Vile, 1999, Anderson, 1991, 1997). There is a continued debate amongst these various schools as to the precise definition of various semiotic concepts. Nevertheless, there are a number of basic assumptions which are common:

• communication, meaning-making and culture are linked through a belief in a shared understanding of signs,
• any semiotic element (sign or text) has at least two planes, an expression plane and a content plane,
• each semiotic element may be read in any number ways other than that intended by the author,
• texts and signs cannot be contextually neutral and offer statements at a level of power, solidarity, and ideology,
• signs are not static, their meaning and their place in meaning making changes for individuals through time.

Computer-based signs in this context includes textual cues, images, icons, and sounds. Semiotics does not recognise that any particular sign is truly ‘universal’. It all rather depends on context: both local and global. Studies have tried to define and quantify the difference between a sign and its meaning (Blankenberger and Hahn, 1991) without much success. This ‘failure’ is not however surprising to semioticians, since it is due to a fundamental semiotic principle: that both the context of the sign and the interpretant of the sign alter the meaning of the sign itself.

Figure 1: Semiotics, SMDF and cross-cultural web-site design

Our Shared Meanings Design Framework (SMDF) aims to bridge the gap between the semiotic paradigm, which is essentially generic, and web interface design. Indeed, our aim is to enhance web interface design so that it matches internal company culture, as well as being capable of reaching across cultural and international boundaries. Figure 1 above gives a schematic idea of exactly what we are proposing in conceptual format.

SMDF places considerable emphasis on HCI as the crucial mission critical area, where issues of meaning, complexity and usability are central. End user received meaning is dependent on a number of interrelated semiotic layers. These layers are shown in Figure 2, and the SMDF semiotic enhanced systems life cycle model shown in Figure 3.

Figure 2: End-user received meaning dependencies & semiotic layers

end user:

meanings

Our SMDF framework consists of a carefully sequenced application of various semiotic enhanced (or semiotically focussed) techniques; our aim being to underpin web-site development from initial requirements elicitation through to post-implementation and review. The remainder of this paper seeks to explicate Stages 1-3 in some detail. Stages 4 and 5 are not described, partly due to space considerations, and partly as these are currently being debated by the authors and by the growing international SMDF community (the interested reader is referred to http://www.mailbase.ac.uk/lists/smdf/ or to http://smdf.org for details of on-going dynamic development of SMDF).

Figure 3: Overview of SMDF Methodology: stage by stage

Much work has been carried out in recent years with respect to formalising and supporting the crucial elicitation phase of information systems development. It has become increasingly clear that traditional ‘hard’ system approaches to requirements elicitation, where often the focus is on the rigorous construction of a formal requirements catalogue and its subsequent iterative refinement, are often failing to deliver the goods. E-commerce systems are making increasing demands upon on all those involved to deliver credible front-end components which engender trust and confidence in both users and company stakeholders alike. Furthermore, the highly distributed and heterogeneous nature of the intended user population of the Internet makes it difficult to empirically validate the usability of the final artefact using a group of users. Such approaches assume that we can assemble a user group and measure user performance objectively, using a standard set of so-called usability metrics whilst also defining in the process, the goals, tasks and environmental context within which the system is to be used. Many web interfaces with a potentially global and thus intrinsically cross-cultural audience cannot easily be replicated with confidence through the assembly of user groups, without firstly identifying and analysing the shared meanings of the application.

The semiotic aspects of any web-site is vital in building the necessary trust and confidence in the system and are critical to determining the ultimate usability of the system itself. We therefore go on to propose a novel approach in which the shared meanings are negotiated iteratively by all the system stakeholders until ‘agreement’ is reached. We believe that before familiar elicitation techniques such as interviewing, record inspection, and observation are used, we should firstly examine the signified of an organisation. We believe that the adoption of the semiotic paradigm will help to maximise the ‘goodness of fit’ of an e-commerce system to any particular organisational culture. We can conveniently divide the signified into several levels as shown below in Figure 4. We can relate these levels of the signified to the specifics of systems, by applying a common sense checklist approach well before traditional system elicitation techniques are used. Indeed, the results of the checklist may well influence the optimal choice of elicitation method(s) themselves. A suggested checklist is included in Figure 4. Additionally we are proposing to enhance and support this process through the development of a semiotic engine designed to explicate and elicit shared-meanings ‘quanta’ of requirements early in the development of a web site. Our semiotic engine is at present at a very early stage of development.

Figure 4: The Signified and Exemplar semiosis Checklist

Thus, the starting point can be seen to be intra-cultural organsiational stakeholder semiosis. A site must in our view seek to address intra-cultural issues, to identify any internal trans-national aspects of its own organisation, and to seek to ensure a good match between the intended system and that organsational culture, before addressing external cross-cultural issues. Organisations which have successfully met this challenge seem to have gained tangible benefits (Thornbury, 1999).

Semiotics provides us with a useful analytical framework within which notions about the cultural and social context of HCI can be approached (De Souza, 1993). From a semiotic viewpoint, an interface is essentially a self-referential symbolic sign system that is ultimately decoded (or rather interpreted) by a set of users. Semiotic analysis attempts for example, to separate out the following features of an interface

We believe that semiotics offers the cross-national web-site interface developer two related approaches:

• contextual HCI semiotics;
• web interface semiotic guidelines;

Semiotics can help us interpret the results of various studies that have revealed that an intimate link exists between an interface and its work-setting (so-called ‘socio-cultural’ studies). Semiotics recognises that for meanings embedded in an interface to be fully shared, it is necessary that we firstly analyse the context within which the system is to be used in a systematic fashion. What semiotics offers here is simply a unifying coherent analytic standpoint in which users are seen to act as interpretants. What we are seeking to interpret is the object represented by the sign. We would argue strongly that in order to interpret computer-based signs successfully, users often draw their cues from the contextual features present in their local environment (Andersen, 1991). We call this contextual HCI semiotics. We can seek to capture the following local semiotic contextual features by asking the following types of questions.

Semiotics deals with meanings associated to signs. These include all types of signs and signification including language, pictures, symbols, actions etc. Particularly from the customer's perspective any Website is clearly the sole basis upon which decisions about the status, authenticity, and trustworthiness of others is made. We are in the process of evolving a set of cross-cultural guidelines.

We feel that HCI semiotics and state modelling fit particularly well together, since they can be used together to model and control both surface level complexity and hence, the semantic complexity of an interface. Formal methods have their own strengths, however we feel that our own approach offers greater simplicity and easier readability potential for e-commerce. Our initial response to controlling surface level complexity, and hence maximise transmission of intended meanings, has been to develop our own visual formalism, (French, 1997). We have chosen to adopt our own notation based around the familiar notion of a state transition diagram.

Figure 5 Home page of Etrade.com

Semiotics, and semiotic design give a perspective from which to view the entire e-finance experience. Clearly architecture or strategy alone will not encourage customers, but by the same token a non-functional semiotically rich web-site will not encourage customers to return.

Figure 6 Barclays-stockbrokers front page

Our second example shows how state modelling can help in the analysis of complexity in a web-site dialogue in terms of user and system states and actions. It illustrates SMDF Phase 3. In a Taiwanese home shopping site (Figure 7) we deliberately choose a non-English (and non-western) culturally based site so as to emphasis the potential ‘universality’ of the technique across cultural boundaries. We provide an English language textual commentary to aid decoding of meanings.

The semiotic environment is complex for a number of reasons. Firstly some signs use traditional mandarin characters (thereby potentially excluding both users from mainland China) and non-mandarin speakers; whereas some signs use English textual signs. A visually complex mix of icons and various heterogeneous image types creates complex syntagmatic chains. A clear cue-card metaphor is employed: but visually we do not know how best to ‘scan’ it to extract meanings (horizontal, vertical, l-right, r-left).

We give below in Figure 8 an exemplar transaction specification within SMDF Phase 3 (booking a flight from New York to Taipei), to show how our notations can enable a designer or user to visualise how complex a transaction is. The intended user of this site is anyone wishing to purchase a ticket from N. America to the rest of the world – hence the web-site is not primarily intended for native mandarin speakers alone but is truly international.

Space precludes a detailed analysis. However it is clear that the logical dialogue of booking a ticket is clear and linear. However the semiotic environment is inconsistent particularly in its mix of different languages, icons and images. For example the ‘security’ aspects are only in English text format, whereas the day / time flight selection screens only in Mandarin (with no iconic or English cues). This confusing mix lead one to doubt whether the site is fully accessible despite the clear state-model which can generated from Figure 8.

In order for our approach to be more widely adopted several barriers must be overcome:

We hope that our paper has gone some way towards addressing the first two of these. We now expect to begin the task of empirically validating our approach on a wider basis that hitherto. We acknowledge however, that the task ahead is not going to be an easy one. Partly, this is due to the conflicting ideas of semioticians themselves and partly due to scepticism amongst those in the HCI community concerning the value of the semiotic paradigm. We hope we have made a start and during the course of our paper and demonstrated that the semiotic paradigm, as exemplified and interpreted by our SMDF methodology can indeed help us meet the difficult challenge of web design.

Figure 8 Exemplar transaction