Conversation Theory

Map Location
D, 2, 3
Thread Locations
Pages 23, 31

R. David Lankes

Agreement Description

In an interview with an Italian library journal, I was asked,

“The conversation” is a brilliant metaphor, since The Cluetrain Manifesto, but somehow vague. According to you, which forms can the conversation take?

I responded that conversation is not a metaphor. When I say “knowledge is created through conversation,” I mean that at least two parties are actively going back and forth in an engaged manner and language is being exchanged. Why “parties” and not people? What do I mean by the notion that language is exchanged? Let me start at a basic level. This is all grounded in Pask’s Conversation Theory (although he does it at a much deeper level).

A conversation has four parts:

  1. Conversants: at least two parties,
  2. Language: sets of meaning going back and forth,
  3. Agreements: shared understandings between the conversants arrived at through language, and
  4. An Entailment Mesh: a collection and relation of the agreements.

Conversants: Parties to a Conversation

In a conversation, you have at least two parties or “agents.” Why not call them people? Because agents are a scalable notion, that is to say, it can be two people (you and I), two groups (say a teacher and his or her students), two organizations (like a library and a vendor negotiating a contract), two countries (a treaty), or even two societies (the great conversation on the meaning of life).

Likewise, these agents can be within a single person. In fact, it is the basis of a lot of instruction and education theory. Call it metacognition or critical thinking skills, or simply arguing with yourself, you have these conversations all the time. If you just asked yourself, “What does he mean by that?!,” who are you asking? Pask, in about 100 pages of dense prose, says that you are in a conversation with different aspects of yourself set up to come to some agreement about a concept. Waking up in the morning and deciding what to wear (“This makes me look fat, this is too dressy…”) is a conversation.

Also, this conversation can happen over a great period of time and through a series of media. So, you read this book (my part of the conversation), think about it, and send me an e-mail about how you agree, or disagree, or simply want clarification. That is a conversation (to be precise, if you are reading this book, you are having a conversation with yourself; it is only if you start sending me feedback that I get involved).

Language: Talking at Their Level

So what are these two or more agents doing? They are exchanging language. This may seem obvious, but it has a lot of implications. There is a large body of research about how people exchange language. For example, there is a discussion on how people know how to take turns in a conversation. There is active research in how people determine things such as power relationships in a conversation (e.g., who is in charge). All of this research is relevant here and can be (and has been) applied to conversations in libraries.

However, Conversation Theory does not directly examine these aspects of conversations. It omits them for several reasons, not the least of which is that much of Pask’s work predates discourse theory. More important for us, however, is that Pask wasn’t just dealing with language between two people. Conversations can also be between two organizations or two parts of one individual. So he approached it in a much more general way. It is also a way that has great implications for how libraries work on a day-to-day basis.

When two agents are sharing language, they do so at one of two levels. The first level is directed and pretty low level. This kind of language is used when at least one of the conversants doesn’t know much about a topic. Take, for example, a librarian walking a patron through a database. It might be a simple set of exchanges, such as “click here, then click here, then type in your query.” The patron’s part of the conversation may be a series of clicks and “OK, now what?” Because the patron doesn’t have much knowledge about what he or she is doing, the language is not rich. Pask calls this kind of language L0. L0 language is used to set up the conversation.

In contrast, L1 is language exchanged between two agents when they both share knowledge of a topic (or domain as Pask would call it). In these exchanges, language is used to build a common understanding of a domain, mutually expand domain knowledge, or clarify some part of the domain. Imagine two librarians debating the finer points of classification, for example.

It is easy to say the patron/librarian example from before was an example of bad instruction (“Don’t just teach where to click, teach why”), but that is a bit simplistic. Good instruction (and Pask would argue, good systems) attempt to raise the conversation from L0 to L1.

However, you have to start somewhere. Furthermore, there are many times when L0 is just fine. Think driving directions (I just want to get to the library. I don’t need a lesson in city planning).

Agreements: Agreeing and Agreeing Not to Agree

So we have a conversation where information is being exchanged in a sequence between two agents (people, organizations, countries). That alone does not create knowledge; that is simply a process between two black boxes.

If a conversation is at its heart simply a back and forth exchange, what’s the point? For example, I could say “1,” you could say “2,” then I say “3,” and so on. Are we really learning anything? Not really because we already know what we are doing (counting up by one), and that means we already know this. No, to learn something, we must seek agreements. That is, we go back and forth making an assertion and seeing what the response is. If I say something and get back an unexpected answer, I need to figure out why. So if I say “1” and you say “3,” I might ask, “Are we counting up by 2s?,” and you might say,”Yes.” I might then say “fine” (meaning I have learned what our task is) or even “5,” and we can continue. This is a pretty minor case of learning, I grant you.

Once we’ve been doing this for a while, we build a whole host of agreements, on which we can seek new ones in new conversations. So if I say “libraries are cool,” for that to mean anything to you somewhere in the past you had to have had a conversation on what the words “library” and “cool” mean. If not, I could just as well have said that we need to turn up the heat in libraries, or, to a C programmer, sets of precompiled functions I can include in my software are really neat. In truth, this book and your internal and external debates are seeking agreement on what we mean by the word “librarian.”

Already you can glimpse some of the implications of Conversation Theory. There are many libraries that claim an educational mission. If learning is an active set of agreements and conversations, then simply providing access to information is insufficient to fulfill our mission! Acquiring materials, organizing materials, and presenting materials may aid in conversations, but they are insufficient to educate. We must present a forum, tools, and opportunities for agreement and conversation. Further, librarians need to actively engage communities in seeking agreements.

So, we have a process of information interchange in sequence that seeks out agreements. It should be noted that an agreement can include “We will never agree on this.” The collection of these agreements is kept in a memory that can be represented in something Pask calls an entailment mesh.

Memory and Entailment Meshes: Pask’s Tangles

Your memory constitutes what you know about the world. It is the sum of agreements we retain after our conversations. However, it is not a simple list or blob of these agreements. As we just discussed, agreements build on themselves. Memory is the agreements and the relationship of these agreements. If you want to impress your friends, you could call memory a “knowledge representation” system like ontologies and semantic networks. If you have no friends, think of it like a map of the stuff you know.

The important thing you need to know is that your memory is all about relationships. That is to say, Conversation Theory talks about the fact that you understand the world not as a series of isolated events or facts but as a dynamic network of agreements and understandings. So to “learn” (Pask talks about “knowing”), a person needs to actively relate new information to what is already known.

Because we’re in the thick of theory, I have to introduce another important phrase: entailment mesh. An entailment mesh is a method of representing the relational nature of one’s memory. It is related to a whole host of visualizations like concept maps or brain maps. Why bother with a fancy phrase? Well, for one thing, to keep us in line with Conversation Theory, but more important, for precision. What’s in your head stays there. Although there is a lot of good research from cognitive psychology to neuroscience that attempts to understand precisely how memories are stored in the brain, Conversation Theory does not. It is much more concerned with how these relationships are expressed, particularly in analog and digital systems. Whenever you create a representation of the memory (try to show how things are connected), you are creating an entailment mesh.

Related Artifacts

Wikipedia entry on Conversation Theory:

Todd Marshall has put together this bibliography on Conversation Theory and the works of Gordon Pask:

Bechtel, J. M. (1986). Conversation, a new paradigm for librarianship? College & Research Libraries, 47(3), 219–224.

Bernard, S. (1980). The cybernetics of Gordon Pask. International Cybernetics Newsletter, 17, 327–336.

Fisher, K. M. (2001). Overview of knowledge mapping. Mapping Biology Knowledge, pp. 5–23.

Ford, N. (2004). Modeling cognitive processes in information seeking: From Popper to Pask. Journal of the American Society for Information Science and Technology, 55(9), 769–782.

Ford, N. (2005). “Conversational” information systems: Extending educational informatics support for the web-based learner. Journal of Documentation, 61(3), 362–384.

Glanville, R. (1993). Pask: A slight primer. Systems Research, 10(3), 213–218.

Lankes, R. D., Silverstein, J. L., & Nicholson, S. (2007). Participatory networks: The library as conversation. Technical Report. Information Institute of Syracuse, Syracuse, NY.

Lankes, R. D., Silverstein, J. L., Nicholson, S., & Marshall, T. (2007). Participatory networks: The library as conversation. Information Research, 12(4). Retrieved from

Laurillard, D. (1999). A conversational framework for individual learning applied to the “learning organization” and the “learning society.” Systems Research and Behavioral Science, 16(2), 113–122.

McKeen, J., Guimaraes, T., & Wetherbe, J. (1994). The relationship between user participation and user satisfaction: An investigation of four contingency factors. MIS Quarterly, 18(4), 427–451.

Pask, G. (1975). Conversation, cognition and learning: A cybernetic theory and methodology. Amsterdam: Elsevier.

Pask, G. (1996). Heinz von Foerster’s self organization, the progenitor of conversation and interaction theories. Systems Research, 13(3), 349–362.

Patel, A., Kinshuk, & Russell, D. (2002). Implementing cognitive apprenticeship and conversation theory in interactive web-based learning systems. Sixth Multi-Conference on Systemics, Cybernetics and Informatics. International Institute of Informatics and Systemics, 523–528.

Pimentel, D. M. (2007). Exploring classification as conversation. In Tennis, J. T., Eds., Proceedings North American Symposium on Knowledge Organization, 1, 1–8, Toronto, Ontario. Retrieved at

Scott, B. (1993). Working with Gordon: Developing and applying conversation theory (1968–1978). Systems Research, 10(3), 167–182.

Scott, B. (2001). Cybernetics and the social sciences. Systems Research and Behavioral Science, 18(5), 411–420.

Thomas, L., & and Harri-Augstein, S. (1993). Gordon Pask at Brunel: A continuing conversation about conversations. Systems Research, 10(3), 183–192.

Wenger, E. (1998). Communities of practice: Learning, meaning and identity. New York: Cambridge University Press.

  • Pingback: A Conversation About Knowledge Creation « Tiny Glass Houses()

  • Pingback: Libraries are not in the construction business « Sense & Reference()

  • Pingback: » Ch-ch-ch-changes Keep Calm (and ask a Librarian)()

  • Pingback: New Librarianship and open questions | Sense & Reference()

  • Craig Patterson

    Hi Richard, I’m not clear on how two parties can exist within one person – it sounds like split personalities to me. “…at least two parties are actively going back and forth in an engaged manner and language is being exchanged.” I’m not clear where this second party in my head comes from, but I’m keen to attend because I like parties ;) These different aspects of myself that I set up – where do they come from? Where do they go? Should I offer them tea? It seems like this sort of thinking could be tied up with and dependent upon identity theories. Have I misunderstood something, or am I missing the forest for the trees?

    • rdlankes

      Have you ever struggled over a decision where you have strong feelings both ways? Or asked yourself “what the heck does this mean?” Who were you asking. This internal conversation, in education called critical thinking or metacognition, is the process where we try out ideas and decide whether to incorporate them into what we know. It is not as grand (or as clinical) as a full blown personality. Rather it is a cognitive process where we array contradictory evidence and make decisions.

      • Craig Patterson

        Yes, yes, and I’m not sure. I suppose when I ask myself a question I’m asking my ‘self;’ but isn’t ‘talking to yourself’ a figure of speech? Is it right to go from ‘I am talking to my self’ to claim that two (I & self) parties are involved in a conversation? It seems that internal events can be represented in this way, but I’m not sure it’s necessary.

        Suppose I’m playing chess against a computer and I represent my internal state like this:

        Should I move my knight to e5?
        No, the bishop on b7 will take.
        But then my pawn on c4 will take.
        But what if that weakens the king’s defences?
        That doesn’t matter because I can mate in 3 after that bishop’s gone.

        Fairly simple, but ‘strong’ feelings each way. Has knowledge been created? Sure, I think I can win in three moves, and am going to move my knight. Were two parties involved? It seems it depends on how you describe it (so above you could put alternating labels at the start of each line, or you could say each line originates from the same source), and if you can describe it as a monologue instead of a dialogue, shouldn’t we? So: does Conversation Theory apply to internal states?

  • Ariel Muvhunzwi

    Knowledge creation seems to a complex process. I also agree with you David on how two parties can within one person…. Maybe we can take this from literary work perspective where one communicates with the outside world through writing , artfects or literary works.

  • Lis Marrow

    As a school librarian, conversations are what we do – the interesting things I have got from this module though are the dialogues we have with ourselves and also how we can help the students to find what they really need….how they can learn with us rather than from us…if I have got it right, I would say that Learning is happening when everyone in the conversation has gained an insight of some kind….

  • geotechlibrarian

    My understanding of this theory in practice is: to learn something, we must seek agreements.Moving from our dispensing of knowledge, to facilitating learning. It’s about conversation, and dialogue, meaning it is not one way, but circular;

  • Pingback: Well, Wadda Ya Know? | Chad On Data()