HOPE - Home   |  The Academy   |  Cyber Library   |  Art Gallery   |  Comments   |  E-mail


Understanding Organizations

Floris and Winograd



11.5  Tools for Conversation

 We have been arguing that the domain of 'decision support' — dealing with alternatives, valuations, and choice — is not the most promising domain in which to build computer tools for managing.  In this section, we will outline another kind of computer tool, based on the theory of management and conversation we have been developing.  We begin by reviewing the basic points of the theory, as developed in this chapter and in Chapter 5:

  1. Organizations exist as networks of directives and commissives.  Directives include orders, requests, consultations, and offers; commissives include promises, acceptances, and denials.

  2. Breakdowns will inevitably occur and the organization  needs to be prepared.  In coping with breakdowns, new   networks of directives and commissives are triggered.

  3. People in an organization (including, but not limited to managers) issue utterances, by speaking or writing, to develop the conversations required in the organizational network.  They participate in the creation and maintenance of a process of communication.  At the core of this process is the performance of linguistic acts that bring forth different kinds of commitments.

 In fulfilling an organization's external commitments, its personnel are involved in a network of conversations.  This network includes requests and promises to fulfill commitments, and reports on the conditions of fulfillment of commitments, reports on external circumstances, declarations of new policies, and so on.  The organization encounters requests and other external contingencies that it can deal with by making commitments that can be fulfilled by the activation of certain special networks of recurrent conversations, where only certain details of the content of the conversations differ, not their general structure.  These networks of recurrent conversations are the core of the organization; they are embodied as intercommunicating offices, each specializing in fulfilling certain kinds of commitments.

 A person working within an organization is always concerned with questions such as 'What is missing?', 'What needs to be done?', and 'Where do I stand in terms of my obligations and opportunities?'.  In situations where many people must act together, the problem of coordination becomes a crucial one.  For many organizations, it is a matter of survival.  The networks of commitments and the conversations in which people participate are becoming larger and more complex, and the complexity of organizations has gone beyond  the point where it can be controlled without appropriate tools.

 The conversation network can be given firmer support with the help of appropriately designed computer-communication technology.  The point of applying computer technology is to help anticipate and avoid breakdowns.  It is impossible to completely avoid breakdowns by design, since it is in the nature of any design process that it must select a finite set of anticipations from the situation, But we can partially anticipate situations where breakdowns are likely to occur by noting their recurrence, and can provide people with the tools and procedures they need to cope with them.  Moreover, new conversational networks can be designed that give the organization the ability to recognize and realize new possibilities.

 Many systems designed by computer professionals are intended to facilitate the activity of an individual working alone.  Although such tools (including word processors, filing systems, program creation aids, etc.) are useful, they leave out the essential dimension of collective work.  In a few work environments, the coordination of action is of secondary importance, but in most it is central.  It is possible to develop computer-based tools that can be used in requesting, creating, and monitoring commitments.  They provide a relevant answer to the question 'What do I need to do?', or, as we prefer to state it, 'What are the states of my active commitments?'.

 The conversational dimension permeates every realm or coordinated activity, whether it be computer programming, medical care, or selling shoes.  The details differ in different work settings, but there is a common basis of theory and a common regular structure.  The rules of conversation are not arbitrary conventions like the rules of chess, but reflect the basic nature of human language and action.  The taxonomy of speech acts and the conversation networks presented in Chapter 5 deal with the fundamental ontology of linguistic acts, and are not based on a particular culture or language.  They are central to the design of tools that operate in a linguistic domain.

 We are not proposing that a computer can 'understand' the speech acts in a particular utterance.  It is impossible to formulate a precise correspondence between the conventions of the written word and the structure of the commitments in a conversation.  What we propose is to make the user aware of this structure and to provide tools for working with it directly.  This is being done experimentally in a computer program that we are developing called a 'coordinator', designed for constructing and controlling conversation networks in large-scale distributed electronic communications systems.  A coordinator is part of a 'communication network' (which might be based on local networks, time-sharing or advanced telephone exchanges) to which all of the participants have access through some kid of 'workstation'.  Its objective is to make the interactions transparent — to provide a ready-to-hand tool that operates in the domain of breakdown generated by conversations for action.

 There are a surprisingly few basic conversational building-blocks (such as request/promise, offer/acceptance, and report/acknowledgment), which frequently recur in networks.  This means that the technological capability that we need to develop to offer support systems for conversational networks can be modular in design.  What is important here is to intuitively grasp that the development of a conversation requires a certain finite selection of possibilities that are defined by the opening directive and the conduct of the speaker.  It is like a dance, giving some initiative to each partner in a specific sequence.

 In addition to facilitating linguistic action, the computer can deal directly with the temporality that is central to the network of commitment.  A promise is not really a promise unless there is a mutually understood (explicitly or implicitly) time for satisfaction.  More subtly, a request is not fully formed unless there is a time specified for reply.  In unstructured social settings, these time conditions are understood by the participants through their shared background, and may never be made explicit.  In structured organizations they are stated directly, and in cases like contract negotiations they are dealt with in laws.  Time is not an incidental condition, but a critical aspect of every speech act.  Breakdowns in conversation for action arise whenever something fails to happen on time.  This in turn triggers other conversations to create new conditions of satisfaction, or to terminate the conversation without satisfying them.  The computer can keep track of time relationships within the network and use them to help anticipate and cope with breakdowns.

A coordinator supports a number of operations:
Speech act origination  An individual performs a speech act using the coordinator by selecting the illocutionary force from a theory-based set of alternatives, indicating the propositional content in the text and explicitly entering temporal relationships to other (past and anticipated) acts.  In addition to having a direct specification of its force, a speech act is related to others, for example as the response to a request, or as a request being made in order to satisfy some previous commitment.  These relationships are made explicit, in the way a speech act is entered into the workstation.  The need to select among pre-structured alternatives for possible illocutionary forces serves as a kind of 'coaching' that reveals the space of possibilities and the structure of different acts within it.

Monitoring completion  Much of the moment-by-moment concern of language is directed towards the completion of conversations for action.  The questions 'What do I have to do now?' and 'What do I need to check up on?' are really questions about the movement of a conversation towards a state of completion (which may or may not include satisfaction of the initial request), as described in Chapter 5.  The coordinator can keep track of where things stand and when they will change.  This can be used to generate reminders and alerts, and to provide a clear picture of what is happening and where potential breakdowns lie ahead.

Examination of the network  An individual can display part of the conversation network, showing the conversations and their status, the individual commitments and requests, and their relationship to others.  It is possible, for example, to find out what requests were generated in anticipation of breakdown in satisfying a particular commitment, or what requests are still awaiting a response from a particular individual.  The details of the interaction (for example using a graphic display) are important to making the tool ready-to-hand, but theoretically central.  The key is that the network is observed in the space generated by the structure of conversation.

Automated application of recurrence  Every organization includes situations that recur and are handled in a standard way.  For example, if a certain request (e.g., for payment) has not been met within a certain time, other requests are made (to the same party or others).  The coordinator can be given this pattern and trigger the corresponding acts without direct intervention.  Of course in doing this, it is important to remember that it is never the computer that makes a request or commitment.  A person can specify a recurrent request or commitment, instances of which are generated automatically.

Recurrence of propositional content  So far we have not talked about the propositional content of the speech acts.  This is an intentional strategy, in that the crucial dimension of conversation for the coordinator is the illocutionary content and its attendant temporal relations.  But of course there a re recurrences of propositional content as well.  A 'purchase order' or 'travel advance request' or any other such form is designed to facilitate the generation of a request or commitment dealing with a particular content.  The creation and use of 'forms' of this kind can be integrated into a coordinator, within the framework provided by the basic conversational, as illustrated in Chapter 12.

 In using a coordinator, the individual is faced with a restricted set of possibilities.  It is not the same as a face to face conversation, a telephone call, or even an electronic message.  Because the illocutionary forces and temporality are specified explicitly, it is necessary to be conscious of them and to have a mutually visible manifestation of them.  This is valuable in a wide variety of everyday communications within organizations, but it is not a universal communicating device, equally applicable to all situations.  In many contexts this kind of explicitness is not called for, or may even be detrimental.  Language cannot be reduced to a representation of speech acts.  The coordinator deals with one dimension of language structure — one that is systematic and crucial for the coordination of action, but that is part of the larger and ultimately open-ended domain of interpretation.

 One existing coordinator system (Cashman) is part of a distributed programming environment for maintaining large software collections.  As programs in this collection were distributed and used, people needed to fix bugs, make improvements, and produce updated versions.  This process was extremely difficult to manage, leading to long delays and failure to meet the needs of software users.  By providing a computer tool to maintain structure of the requests and commitments, they were able to greatly improve productivity.  Other such systems are in development at a number of institutions.

 We conclude here by pointing out that computer tools themselves are only one part of the picture.  The gain from applying conversation theory in organizations has to do with developing the communicative competence, norms and rules for the organization, including the training to develop the appropriate understanding.  This includes the proper terminology, skills and procedures to recognize what is missing, deteriorated or obtruding (i.e., what is broken-down) and how to cope with the situation.

 People have experience in everyday dealing with others and with situations.  Nevertheless, there are different levels of competence.  Competence here does not mean correct grammatical usage or diction, but successful dealing with the world, good managerial abilities, and responsibility and care for others.  Communicative competence means the capacity to express one's intentions and take responsibilities in the networks of commitments that utterances and their interpretations bring to the world.  In their day-to-day being, people are generally not aware of what they are doing.  They are simply working, speaking, etc., more or less blind to the pervasiveness of the essential dimensions of commitment.  Consequently, there exists a domain for education for communicative competence, as a field concerned with certain general relationships between language and successful action.

 The process of communication should be designed to bring with it a major awareness concerning the occurrence of breakdowns and the directives appropriate to them, as well as the occurrence of commitments.  People's conscious knowledge of their participation in the network of commitment can be reinforced and developed, improving their capacity to act in the domain of language.


HOPE Home Page Academy of Jerusalem Cyber Library Comments Form E-mail