The implications of manipulable inter-medium encodings for coaching and learner modelling

Domains typically have associated with them some notation, or more generally, encodings formed from elements of a medium, which are used to convey information about the concepts, objects and processes of the domain. Often, entities in a domain can be communicated more directly by encodings in an alternative medium, e.g. diagrams. In some systems, the representative encoding which is available to the user (the external encoding) can be manipulated by the user in ways which are constrained to have meaning in encodings in another medium (the internal, or base medium). Such interfaces embody what is called a manipulable inter-medium encoding (MIME). The ITS implications of such systems have been investigated in studies based on a simple `direct manipulation´ graphical interface (called HUGH). The task was the proof of a theorem of context free languages by constrained exploration of derivation tree diagrams. Fifteen postgraduate researchers took part in the experiment. The HUGH system was not an ITS, but an ITS was simulated by the presence of an observer providing coaching as required. At certain points in the interaction, learners were presented with questions (and corresponding answer windows) designed to elicit their semantic accounts of the operations performed. The study indicated some ITS implications of the use of MIMEs