A Technical Memo
Stephen W. Draper
Department of Psychology
University of Glasgow
Glasgow G12 8QQ U.K.
WWW URL: http://www.psy.gla.ac.uk/~steve
In examining the scenarios, some questions to consider are:
On this view, simply recording such dialogues and getting new learners to hear/read them is enough. This is essentially the position of "lurkers" on email discussion lists, who frequently testify that this is valuable. No indexing is wanted: this learning happens without the learner being driven by their own questions: they learn by having unexpected material thrust on them.
A horrid thought: perhaps the AG software will never be that good at automatically finding the right piece, so perhaps we need to aim at something like telephone directory enquiries: the human (tutor) elicits the learner need in real dialogue; then does a command selecting the answer from store, which is then transmitted without the tutor. The benefit would be to allow the tutor to dispatch more good answers per minute.
Is this in fact the main function of tutors?: they have a lot of stored answers, and the human is mainly needed because a) these are not stored b) there is no software or index system to retrieve them satisfactorily for learners. Only occasionally are human tutors required for truly intelligent feedback.
If this is right, then in general our business is a) capture material as/when used e.g. all replies to student queries. b) organise for fast re-transmission.
This is a general requirement: a) any linear medium cf. students marking textbooks to find the bits they think are crucial, especially in novels which unlike textbooks do not have indexes. b) It is an example of having an index built for (and perhaps by) the learners. c) it is a prime example of a pre-existing external document, with the AG software acting only to point into it.
There are cases where users will never read a primary exposition, but always go in question-driven mode e.g. reference manuals, university library Q&A. But actually, can still regard the manual as the primary exposition, and an AG as a way of discovering how to write the index correctly i.e. in a user-centered (task-organised) way.
Minimal Manuals do user task analysis to design the structure/index; an alternative is to study help desk interactions to discover the questions/problems users have and later use that to design the ref. manual. The motivation is similar to that of Minimal Manuals: learners/users who won't stop to learn before acting, but are driven by tasks. Here we extend that a bit to users whose goal is learning rather than a material task, but who nevertheless are too impatient to read the given text before they see how it is useful to them.
This idea is related to:
A quite different angle: displayed dialogue is a presentational form, useful pedagogically for its own sake. E.g. Plato's dialogues; many TV shows / documentaries. When is this the best form? Can captured dialogues ever really be it? Are their ideal(ised) learners to play a part in such dialogues?
Dialectic. Looking at multiple claims, counter-claims, resolutions. New Scientist's "The last word" column. [e.g. How do robins make such a loud noise, compared to a radio's loud speaker?]
This is a bridge between index and questions, as the idea here is that the same old content is to be conveyed, but the chosen format is that of Q&A. So perhaps this is basically a way of structuring (indexing) material, so it is that index or structure that is the specific content. However that index IS questions (that students want to ask). They are typically connections to prior experience and conceptions.
The difference between this and the previous item is that here the learner probably reads the Q&A sequentially, as designed by the author; and the questions are put by the author (not the learner), and must be explained and made convincing to the learner. Whereas in a reference manual, the questions only have to be recognised by the learner; and the learner will not look at any material they haven't already thought about as a question for them.
So here, the AG is only (re)used by the author (not by learners); and for them, the important content is primarily the questions (what students didn't understand without asking), though secondarily the answers that then worked to satisfy the students.
In other words, AG is only a second best substitute for a proper primary exposition for the topic. It is worth it in HE if/while primary materials are not debugged, or some other function is being served. More and more I see it as a temporary stage before creating or re-writing the inadequate primary source: the textbook or lecture, or the minimal manual if the domain is help with software.
In fact, it should be seen as an authoring tool for creating primary expositions, given only bugs or questions or a discussion. In a topic that has no proper primary exposition yet (e.g. ACK's design of MM systems; ... or this: the use and function of AG), then you could begin that exposition by either imagining the questions learners will have, or jotting down unrelated points as separate "answers" then authoring the structure. In fact this is just hypertext authoring; or the (usually informal) process of writing a paper from a set of points in the head, then do the structure, then rewrite each point. This is only one step on from artificially creating a discussion.