Notes on: Pask, G.  (1976) 'Conversational Techniques in the Study and Practice of Education'.  In British Journal of Educational Psychology, 46: 12-25

Dave Harris

Conversational learning takes place in 'realistically complex learning materials'.  The process needs to be externalized by providing an apparatus which controls learning and allows records of the steps.  The [computer-based] apparatus should also include subject matter representations, a diagram of the relations between concepts.  These can be connected electronically, so that the learner's steps can be recorded.  There is a reference here to the work of Marton and Saljo [whose contribution appears in the same volume—this volume includes papers from a particular symposium].  Marton and Saljo [M and S] offer work which is 'realistic', but they need 'to redefine common terms (such as understanding) to have a more restricted and precise meaning', and we can also introduce new terms.  The work of other psychologists can also be modified, including Kelly, Laing and Piaget.  What is required is a 'systems approach', where learning takes place through 'interpreted formal relationships', such as 'next', 'adjacent', 'periodic', 'dual', 'sum', and 'product' (14). We can define understanding as a description of the relations between concepts in various topics.  There might be verbal and non verbal descriptions [where the latter include manipulations of apparatus, for example]. Both descriptions and uses of the system are important.

We can see subject matter as a mental map, as in Piaget, but 'entailment structures' would be more formal.  We start with a thesis 'expounded usually by a subject matter expert' (15), which is then broken down into a series of derivations.  We have to introduce 'cyclicity' [closure] to reconstruct the topic and to provide consistency.  The resulting structure is 'merely the expositor's thesis' [after some serious modification, however?].  Entailment involves both 'axiomatic derivations' and 'correspondences, morphisms such as isomorphism' (17).  Entailment of this kind mesh together in a network.  Experts can further modify and ordering this network, for example by saying which topics are peripheral.  There is a similar procedure in Kelly Grids for Pask (18), although descriptors of topics there are personal constructs [but formally exactly the same as expert constructs].  It is possible to use Kelly's technique of repertory grid administration to develop analogies and other correspondences [that is to organize the constructs in grids, possibly laddered together, and to ask the respondent if they apply to other concrete examples—if I remember my Kelly correctly].  However, each topic node must be uniquely identified [for closure purposes].  A class of valid explanations for each topic can then be developed, producing a task structure or 'behaviour graph' (19).  The whole entailment structure, together with its descriptors and behaviour graphs constitute the 'conversational domain'.

Tutorial conversations and transactions on this conversational domain will produce questions aimed at making sense of the topics, or asking for descriptors or explanations.  In this way we can 'explore transactions', or 'aim [a] request', to get to any one topic.  There are only some permitted derivations, however, and these can serve as checks on aims and abilities to get there.  The whole thing depends on some 'higher agreement—an agreement regarding the derivation of the topic' (22), and an agreed form of explanation.  Students also have to agree to the concept being 'stable'[presumably meaning fixed within the network].

The whole thing can be computer controlled, and this is useful in research to standardize experimental conditions.  A network will still provide a large variety of paths and demonstrations, so 'students have considerable freedom to learn within the constraints of the system as a whole' (23).  With a computer in charge, students become their own teacher, setting goals and attempting to understand topics. 

Pask has invented a programme called CASTE, [ discussed in my chapter 2] with a portable version INTUITION, described 23-25, which he uses in his experiments.  He also uses the TEACHBACK procedure [described earlier], where the teacher is neutral, merely asking for descriptions of how tasks are learned and explanations derived.  This is a more rigorous and controllable form of diagnosis rather than, say, diagnostic interviews [presumably a critique of M and S again?].

Back to education studies