I started digging in today into what I’ve found to be a very interesting volume, Questions and Information Systems (1992). (It was somewhat quaint to hear references to ARPANET.)
While several chapters are about what were at the time termed ‘expert systems’ (‘algorithms’ being a term that sounds to me like it has the same level of being in vogue now as it did then), some chapters focus on the perhaps more classic or basic problem of what questioning involves. These certainly draw from the technological frame, but less than they do from (I would say) philosophy and linguistics.
One example is Marianne LaFrance’s chapter, which presents a nice summary of the problems with the ‘usual’ ways of thinking about questions, e.g. as ‘knowledge acquisition’ (p.15); ‘In fact, acquisition of knowhow from an expert is one variation on the venerable question of how any person manages to transfer hard-won personal proficiency to someone who is less skilled’ (p.15).
I was also excited by her six axioms about the ‘process of question asking‘ (p.16), which I found ring truer for their being counter-intuitive:
- Information is not extracted by questioning.
- Questions require common ground.
- All questions are leading questions.
- Questions derive from knowledge rather than ignorance.
- Questions occasion the telling of stories rather than the furnishing of answers.
- Good answers ring true rather than are true.
The other chapter I was studying was Graesser et al’s chapter on ‘Mechanisms that Generate Questions’ (p.167). They proposed a rather comprehensive taxonomy of inquiries, as well as propose a summative framework of question-generation mechanisms. The four main groups of mechanisms they identify are:
- Correction of Knowledge Deficit
- Monitoring Common Ground
- Social Coordination of Action
- Control of Conversation and Attention (p.175)
For my own reference, I’m recording here my own alternative grouping of the 18 categories in their taxonomy of inquiries (pp.172-173).
- I think the first basic group includes verification (is X, y?), comparison (how is X vs. Y), and disjunctive questions (is x or is y?). I imagine these types of questions as operations that localize the exchange within some topic area. One feature I observed of these question types is that the questioner has to supply the range of knowledge to the answerer.
- I thought ‘concept completion’ (Wh-questions) was closely related to the first group, in that the questioner still supplies significant identifying information, although the object/referent of the question is left genuinely open. This distinguishes it slightly from the above group, however, in that it presents the opportunity for a new knowledge-entity to populate the knowledge base.
- Following from this, I thought that ‘definition’ and ‘example’ are species of questions that are intensively focused on populating new areas of the knowledge base.
- ‘Interpretation’, ‘Feature specification’, and ‘Quantification’ all appear related to eliciting the answerer’s assessment of some situation.
- ‘Causal antecedent’, ‘Causal consequence’, and ‘Goal orientation’ seem to be focused on eliciting the answerer’s understanding of a narrative.
- The frame for ‘Enablement’ and ‘Instrumental/Procedural’ questions seems to be a problem-solving frame.
- The last group of questions (‘Judgmental’, ‘Assertion’, ‘Request/Directive’) have in common how little room they leave the answerer to manoeuvre. The question-type listed before these, ‘Expectational’, can be as restrictive as these, but could also occur when the stance involves the questioner genuinely seeking the answerer’s assessment. However, this might be better classified under ‘Interpretation’.