All organizations depend on some form of communication [3]. Communication may be seen as the basis for action, and in some cases, it is the action itself. Communication in organizations can be difficult, however, often leaving people frustrated with communication breakdowns, unfulfilled commitments, and overwhelmed with meaningless messages. And information technology has further complicated communication in two ways. First, organizations have come to rely on technologies such as email, instant messaging, electronic meetings, videoconferencing, voice mail, online transactions and so forth to support their communication. These Communication Support Systems (CSS), in comparison with traditional face-to-face meetings, are less effective for sharing knowledge and building relationships, primarily because they do a poor job of equally distributing and contextualizing information [2]. Secondly, information technology has intensified certain communication difficulties because of the cultural changes it has brought about.
The complexity of implementing communicative action grows with the need for coordination, the contextual demands and the use of scarce resources.
In particular, successful organizations in the information age are characterized by higher amounts of cross-departmental collaboration, diverse thinking, and knowledge sharing [5]. These trends require more complex and more intense communication across different perspectives, based on argumentation rather than rank, and rich with context.
Unfortunately, research addressing these problems has lagged considerably behind the technological advances. There is very little in terms of design guidelines for CSS, let alone a comprehensive view that rests on theoretical grounds. This article presents a framework built on elements of the language-action perspective (LAP) augmented with theories of human behavior and uses it to develop design principles.
Conceptual Basis
Two important foundations of LAP are The Theory of Speech Acts [9] and The Theory of Communicative Action [4]. We borrow from both [7], using the word speech (accompanied by the words speaker and hearer) to denote any form of communication, including non-verbal, visual, digital, and so on.
Figure 1 simplifies parts of Habermas’s view of communicative action relevant to the discussion here. Goal-oriented communication takes place within a context of values, norms, and resources. In order for goals to be achieved, coordination between communicators is necessary, as is their commitment to act appropriately. Rationality and complexity characterize the interaction between goals and action. Communicative action achieves the communicator’s goals through a rational process that is consistent with the prevailing context. The complexity of implementing communicative action grows with the need for coordination, the contextual demands (norms and values) and the use of scarce resources.
Rationality. Goal-oriented communication and rationality underscore two important principles that have to do with context and communication quality. According to Searle, speech acts consist of three components: its propositional content (core message), its context or background information (illocutionary context), and the speaker’s intention (illocutionary force). Intentions are of different forms such as assertions (representing real-world states), directives (attempts to get the hearer to act), and expressions (feelings and attitudes). In Figure 2, the linguistic context is drawn as an expansion to the core message: its role is to support the core message. The intentions, drawn at a higher level, govern the content and its context. To understand how these three components affect communication, we first define effective communication.
According to Habermas, comprehensible, trustworthy, and appropriate communication is at once an act of building mutual understanding and relationships between the communicators. Mutual understanding is necessary for agreement and task accomplishment, and relationships are necessary for gaining commitment between communicators. All too often, CSS ignore one or the other, for example, ignoring relationships in structured transactions and mutual understanding in unstructured IRC.
Principle 1: Design must simultaneously consider enhancing mutual understanding and promoting relationships between communicators.
Complexity. To cope with complexity, effective communicators employ communication strategies that adapt communication according to the speaker’s intentions and to the situation (such as the type of task, the physical and psychological distance between communicators, and the prevailing values and norms). An example of a strategy is contextualization, which is the provision of contextual information to explain a core message. The right amount of contextual information is crucial for effective communication: too little or too much diminishes mutual understanding and thereby performance. Effective communicators further adapt communication by varying the medium they use (face-to-face, instant messaging, or email) and the form of the message (formal versus informal messages, private or public, highly organized versus free-form messages) [10]. In other words, reduced mutual understanding and poorer relationships should be expected when systems constrain the users to certain media and forms of message, for example, enabling asynchronous but not synchronous communication or supporting formal but not informal communication.
One straightforward implication is that design should support adaptive behavior to ensure effective communication. For instance, when communicating with subordinates, the system can ascertain that the message is appropriate according to the organizational norms or alert the sender to sexist language that is culturally unacceptable and damaging to work relationships.
Furthermore, adaptive processes require control, which communicators cannot always exercise effectively. For instance, communicators may over-contextualize, increasing complexity unnecessarily and thereby reducing rather than enhancing mutual understanding. Veteran workers tutoring a novice may explain not only the practice of sealing a contract with a client but also describe in detail how this practice came about and how extant practices were revised to overcome costly mistakes. The newcomer cannot absorb this information because the context is unfamiliar and requires too much attention, at the expense of understanding what to do next. In the case of the newcomer, contextualization results in lower rather than higher understanding.
Increasingly, conference systems allow both synchronous and asynchronous communication. Effective communicators may switch from email to instant messaging (IM) when they sense communication is ineffective and requires immediate explanations to understand a complex message. Systems should not only enable both modes but make it easy to switch from one to another, for example, by listing previous email messages that are relevant as part of the IM display. By the same token, CSS should not only allow both verbal and non-verbal communication but combine them seamlessly.
Principle 2: Design should support adaptive behavior, including the contingent use of alternative communication strategies, alternative message forms, and alternative media.
Using IM, especially when several people are communicating about a dynamic situation (such as collaborating on a rapidly evolving problem), can overload users with messages, causing them to miss messages or feel overwhelmed. Systems that monitor complexity, for example, by measuring high rates of unattended messages, can notify the communicators that, until further notice, new instant messages will be restructured and redirected as an organized email message.
Principle 3: Design should monitor complexity and alert communicators to a high propensity of communication breakdowns.
Levels. LAP draws the designer’s attention to the analysis of multiple levels of communication. For instance, e-commerce can be modeled using several layers, two of which are speech acts (the building blocks) and scenarios (complete user interactions with the system) [11]. Each layer builds on its predecessor. Similarly, in developing guidelines for CSS, we consider multiple levels of communication: starting with the level of the speech act, descending its components of core message, context, and intention, and ascending to the higher levels of complete interactions, including the characterization of the situation (goals, values, and norms).
The assumption that people communicate at multiple levels implies two types of support. First, support communicators in selecting and processing the unique information needed at each level. For example, scenarios must involve sufficient threads of speech acts to avoid premature convergence but this requires the possibility of communicating tentative, incomplete, and non-validated information. Second, support communicators in traveling from one level to another. For example, being able to see a focal speech act within a thread of messages (constituting a scenario) helps move up from the speech act level to the interaction level and back. This is easily accomplished with one central window displaying the current focal message and a linked window listing clickable titles of the threaded messages. More sophisticated techniques are needed to determine that a scenario is incomplete by analyzing the intentions of all the separate speech acts. Current CSS are relatively flat, robbing the communicators of the ability to move from one level to another when they find it advantageous to do so.
Principle 4: Design should support multiple levels of communication and easy travel between levels.
Memory. Communication and memory go hand in hand. Communication is necessary for acquiring knowledge that feeds into memory, and at the same time, stored knowledge is essential for effective communication. Moreover, memory processes and structures shape and, during the conversation, dynamically reshape communication. For example, associative networks in memory expand communication, episodic memories affect the content and style of communication, and failed memory aborts or distorts communication. Clearly, the content and organization of human memory is crucial for communication.
Organizational memory is the collection of distributed mechanisms for applying past knowledge to current action, residing in humans, culture, and information and communication technologies. In order to support communication effectively, it must include the information needed to support adaptive, appropriate, and multi-level communicative action. Wiki technologies rely on culture to dictate appropriate, and trustworthy communicative action. Unfortunately, other CSS must resort to surveillance technologies to detect violent communications.
As noted previously, dialogue in which people’s interpretations evolve over time imposes special requirements on memory: information is owned, associative, indeterminate, emergent, and maintained in mixed forms [1]: provenance and governance of interpretations are explicitly defined so that owners are responsible and recognized for knowledge; users can easily travel between diverse interpretations; the system supports maintenance and analysis of diverse perspectives; the system supports indeterminate, partial, and tentative knowledge; it allows the emergence of new categories, constructs, and levels of abstraction in knowledge; and it supports access to knowledge in different ways, including graphic/textual, visual/audio, verbal/non-verbal, and stills/motion. Common CSS tend to stifle dialogue by remembering only final, validated, and complete information with no room for intermediary stages, and similarly tend to suppress diversity by remembering only agreed-upon interpretations.
Principle 5: Organizational memory should consist of speech act components, situations, norms, and values.
Principle 6: Organizational memory should consist of associative information, accessible through multiple media, and represented in multiple forms, allowing for indeterminate and emergent views as well.
Emerging Applications of the Principles
A few early CSS attempted to capitalize on all speech act components (Principle 5). The pioneering work of Winograd and Flores on Coordinator demonstrates how the intention of each message can be specified, making it possible to analyze its content in light of the intention and to minimize misunderstanding [12]. It also became possible to manage commitments and reveal unfulfilled commitments. Coordinator could thus boost productivity, increase customer satisfaction, and enhance trust. Other useful information in memory includes organizational knowledge such as the relations or distance between communicators to determine the likelihood of misunderstanding due to different professional backgrounds. Differences in professional background can be calculated by explicit information in signature files. Similarly, Web-based systems use collaborative filtering and compute a measure of similarity between communicators’ profiles to decide what information to communicate. More sophisticated procedures rely on knowledge of patterns of personal ontology to compute the linguistic distance between the communicators. Given the CSS can determine this distance, it can detect high communication complexity and highlight the danger of misunderstanding.
Today’s communication should be organized in memory to become the situational context of tomorrow’s communication (Principles 5 and 6). Thus, the key to success of CSS will be the clever organization of extant communication to support future communication. We have technology capable of intelligent categorization of messages into simple organizations according to keywords and network structures, as well as advanced retrieval techniques such as text mining. It is time to apply these technologies systematically to support communication.
The support of specific communication strategies using information in memory is geared toward higher mutual understanding and better relationships (Principles 1 and 2). Contextualization is perhaps the prime meeting point between communication and memory. A knowledge-based mailer called kMail is an example of contextualization in CSS [8]. The system builds links to relevant information automatically by parsing outgoing messages to detect possible information that elaborates the message. The challenge is to deliver context only when needed, in other words to support adaptive behavior when it is effective. To do so, the system must be able to determine when contextualization is needed, for example, by detecting differences in professional perspectives or differences in terminologies used. Figure 3a depicts a message as it is sent and Figure 3b shows the message after it has been parsed and transformed into a contextualized message. In kMail, the different perspectives owned by different communicators are indexed so that people can see a message in light of alternative perspectives (see Figure 3b).
Other advanced representations may be more suitable for communicating abstract perspectives. Spider [1] is designed to present context at several levels of abstraction and in a variety of forms, including cognitive maps that highlight the similarities and differences in the communicators’ perspectives (Principles 2 and 4). Cognitive maps are elaborative representations of factors and relationships drawn as maps consisting of nodes and links, for example, `market share’ increases `profits’. The maps can be expanded into more detailed maps and may be linked to other forms such as spreadsheets, graphs, and free form texts. The representations are organized as hypertext. Without such painstaking efforts to represent and study the partner’s perspective, there is little chance of mutual understanding. Furthermore, computing the distance between cognitive maps allows the system to signal to the user the difficulty in achieving mutual understanding and the need to adapt communication accordingly.
Spider also demonstrates the idea of allowing indeterminate and emergent views in memory to promote dialogue (Principle 6). Imagine Joe has a cognitive map that links two factors but he is not sure that one really leads to the other. He places a 0.5 confidence level on this relationship to reflect indeterminacy. The impact on communication is twofold. First, the message is more accurate. Secondly, the communicator becomes more trustworthy by sincerely revealing hesitations and signaling an openness to listen and reconsider his opinion in light of further dialogue.
EVIDII (Environment for visualizing differences of individual impressions) supports communication between clients and designers to help them in constructing a shared understanding [6]. The assumption is that if the system can identify communication breakdowns due to different languages and different cultures, users will know how to adapt (Principles 2 and 3). EVIDII elicits the communicators’ meaning of terms by visualizing their impressions on coordinated maps. But this is only one perspective and a user can change perspectives from, for example, an image perspective that views the words associated with an image by different people to a person perspective that views the words used by a particular person. Changing perspectives allows the users to examine the information from different aspects but also to obtain a holistic view of the multiple perspectives. This leads to a better appreciation of the values held by communicators that come from a different culture.
EVIDII further claims to boost not only mutual understanding but also relationship by increasing the trustfulness of the communication (Principle 1). Faces of respected communicators that stand behind the language are said to increase trust. Meeting systems that allow even still pictures of the speaker are usually well received by users. Pictures of sufficiently high resolution to signal facial expressions synchronized with the propositional content of the message are beneficial for establishing trust. In EVIDII, synchronous facial and verbal expressions are linked to add credibility.
The six design principles described here emerge from many years of research in CSS that builds on LAP and psychological theories of human communication. Having refined these principles, we can see how they are applied in the systems discussed in this article. However, each system applied only part of the principles. We are lacking a comprehensive approach that takes all six principles into account from the outset and not as an afterthought. Indeed, many of the technologies needed to implement the principles are already available but not utilized because we lack the methodology and the commitment to redesign CSS in the organization. Other technologies must be developed to better identify situations that call for adapting communication and help communicators adapt appropriately. Only then will we be able to fully utilize this crucial resource.
Figures
Figure 1. Simplified view of communicative action (based on [
Figure 2. Three speech act components: core content, context, and intention, which appears on a higher level of communication.
Figure 3. (a) Depiction of a message as it is sent; (b) after it has been parsed and transformed into a contextualized message.
Join the Discussion (0)
Become a Member or Sign In to Post a Comment