Towards higher order questions: varieties of questions

Engaging with Questions of Higher Order: cognitive vigilance required for higher degrees of twistedness (Part #9)

---

[Parts: First | Prev | Next | Last | All] [Links: To-K | From-K | From-Kx | Refs ]

---

The earlier exploration of twistedness led to recognition of domains in which it was fundamental (as with DNA), and where its challenges had been taken up -- whether through helical structural organization or through an understanding of processes organized helically over time. Comprehension of the dynamics is exemplified by the challenges of controlling the three-dimensional movement of a helicopter. The experiential dimensions may be particularly well-recognized by those of more disciplined kinetic intelligence (acrobats, martial arts practitioners, etc). Arthur Young has been most helpful in generalizing the understanding of the requisite cognitive skills in his emphasis on learning-action cycles.

This context leads to the focus of this paper, namely the nature of the questions required to distinguish and engage effectively with the more complex forms of psycho-social "twistedness" that increasingly inform much socio-political action -- for the better or for the worse. It is these questions which may be expected to identify new opportunities and to guide new forms of action -- as well as being inherent to the cognitive vigilance necessary in a turbulent conceptual environment in which most strategic agendas can be framed as "twisted" from another perspective.

Questions in the context of information retrieval: Formally a question may be understood as a speech act of requesting information on the truth value of a set of propositions. This is problematic since the definition is pragmatic rather than syntactic when questions may also be the centre of focal attention in syntax. Furthermore, defining a question in mere formal terms tends to be either too language-specific or redundant.

For John G. Keyes (Using Conceptual Categories of Questions to Measure Differences in Retrieval Performance, 1996):

A question is a formalized request for information. The form of the question communicates the degree of current knowledge towards a given information need. A question such as "Are X and Y related?" implies a different level of knowledge on the part of the asker than does "How are X and Y related?" In the first instance there is uncertainty as to whether the relationship exists, while in the second the relationship is presumed. In this way, the question reflects the cognitive model that the asker has developed in regard to his need for information. Because the cognitive state of the asker dictates the form of the question and also governs the appropriateness of answers for filling his information need, it could be expected that the form of the question would have an effect upon what answers are appropriate and relevant to the information need as stated.

Keyes provides a useful overview of various approaches to developing a typology of questions:

• Lehnert (1977): (1) why questions, (2) how questions, (3) yes or no questions, (4) occurrence questions, which ask about some general act or sequence of acts, and (5) component questions, which are essentially fill-in-the-blank type questions.
• Lehnert (1978): (1) causal antecedent, (2) causal consequent, (3) concept completion, (4) disjunctive, (5) enablement, (6) expectational, (7) feature specification, (8) goal orientation, (9) instrumental/procedural, (10) judgmental, (11) quantification, (12) request, (13) verification
• Graesser and Murachver (1985): sought to identify the knowledge sources and basic processes that contribute to question answering. The first component of this architecture is question identification, which includes identifying question functions and statement elements. They identified seven question functions: (1) WHY, (2) HOW, (3) ENABLE, (4) CONS (i.e., what is the consequence of), (5) WHEN, (6) WHERE, and (7) SIG (i.e., what is the significance of). Each question function is identified with one of three statement elements: (1) intentional actions, (2) events, and (3) states. According to this scheme for example, WHY, WHY, and WHY are different question categories that have different symbolic procedures. In combining question functions and statement elements, they identified a total of 21 categories with different symbolic procedures operating on the knowledge structures.
• Graesser, Person and Huber's taxonomy of question types: (1) Verification: Must a cell have its own security server running? (2) Comparison: How is the cell directory server similar/different to the global directory server? (3) Disjunctive: Does distributing the namespace increase or decrease network traffic? (4) Concept completion: Who invented DCE? (5) Definition: What is a CDS? (6) Example: What is an example of a good cell configuration? (7) Interpretation: Is the network performing well? (8) Feature specification: What sort of server distributes PACs? (9) Quantification: How large is the cell directory service process size? (10) Causal antecedent: What caused my process to fail? (11) Causal consequent: What happens if I give the wrong password? (12) Goal orientation: Why is DEC investing in DCE? (13) Enablement: What do I need to have an operational cell? (14) Instrumental/Procedural: How do I set up the namespace? (15) Expectational: Why doesn't the time server provide an exact time? (16) Judgemental: Will DCE take over the world?
• five conceptual categories of questions based on their semantic representations: (1) Verification, (2) Concept Completion, (3) Causation, (4) Association, and (5) Disassociation *****

Keyes argues that:

The above models all attempt, in one way or another, to capture and categorize the subtler nuances of human communication. If the categories are too broad, then much of the communication will not make sense. Elaboration and subtlety are prerequisites for any system that would attempt to model more fully the human question answering process. In information retrieval, however, the goal is more pragmatic: to improve system performance and to facilitate end-user acquisition of relevant information. Since what is sought is improvement rather than complete viability, one may be able to make progress with a lesser degree of sophistication. Because the user is usually concerned with elaborating a partially known information need, the types of questions asked in information retrieval settings can be viewed as a subset of the types of questions humans asked. In a sense, the user is aware of what he does not know in this setting. Consequently, his questions will reflect this and will be structured in a way that specifies his information need.

Xin Li and Dan Roth (Learning Question Classifiers) haved developed a machine learning approach to question classification, recognizing that locating an answer may depend on filtering out candidate answers based on an understanding and categorization of answer types. For them:

In order to respond correctly to a free form factual ques-tion given a large collection of texts, one needs to un-derstand the question to a level that allows determining some of the constraints the question imposes on a pos-sible answer. These constraints may include a semantic classification of the sought after answer and may even suggest using different strategies when looking for and verifying a candidate answer.

Rhodes' Typology of Questions (1995):

• Content-directed questions (intended to obtain information about subject matter)
• Student-directed questions (intended to obtain personal responses from students)
• Rhetorical questions (self-expression, obtain effect, or a means of giving directions)
• Ambiguous questions (intended to obtain information, but the nature of the question is not clear)

Typology of Questions (A Rossett. Needs Assessment, 1995)

• Problem finding: Is there a problem? What is the nature of the problem?
• Problem selecting: Prioritize identified problem
• Knowledge/skills proving: Ask to perform the task
• Finding feelings: Questions about the feelings and attitudes about the problem
• Cause findings: Questions about the cause of the problems

The 13 conceptual question categories used in Wendy Lehnert's QUALM in which questions can be mapped by an inferential analysis procedure. The taxonomy proposed by Lehnert is primarily based on a theory of memory representation called Conceptual Dependency:

1. Causal Antecedent Why did John go to New York? What resulted in John's leaving? How did the glass break?
2. Goal Orientation For what purposes did John take the book? Why did Mary drop the book? Mary left for what reason?
3. Enablement How was John able to eat? What did John need to do in order to leave?
4. Causal Consequent What happened when John left? What if I don't leave? What did John do after Mary left?
5. Verification Did John leave? Did John anything to keep Mary from leaving? Does John think that Mary left?
6. Disjunctive Was John or Mary here? Is John coming or going?
7. Instrumental/Procedural How did John go to New York? What did John use to eat? How do I get to your house?
8. Concept Completion What did John eat? Who gave Mary the book? When did John leave Paris?
9. Expectational Why didn't John go to New York? Why isn't John eating?
10. Judgmental What should John do to keep Mary from leaving? What should John do now?
11. Quantification How many people are there? How ill was John? How many dogs does John have?
12. Feature Specification What color are John's eyes? What breed of dog is Rover? How much does that rug cost?
13. Request Would you pass the salt? Can you get me my coat? Will you take out the garbage?

Arthur Graesser's Taxonomy of Inquiries has foundations both in theory and in empirical research. It uses Lehnert's 13 categories to which have been added:: a "comparison" category, a "definition" category, an "example" category, and an "interpretation" category. For all the categories in the taxonomy, Graesser conducted a study of empirical completeness, showing that the folowing taxonomy is able to accommodate virtually all inquiries that occur in a discourse:

1. Verification Is a fact true? Did an event occur? Is an F-test a type of statistic? Did it rain yesterday?
2. Comparison How is X similar to Y? How is X different from Y? In what way is Florida similar to China? How is an F-test different from a t-test?
3. Disjunctive Is X or Y the case? Is X, Y, or Z the case? Do the mountains increase or decrease the rain in Oregon? Did he order chicken, beef, lamb of fish?
4. Concept completion Who? What? When? Where? What is the referent of a noun argument slot? Where are the large population densities in North America? Who wrote the song? What did the child steal?
5. Definition What does X mean? What is the superordinate category and some properties of X? What is a factorial design? What does interaction mean?
6. Example What is an example of X? What is a particular instance of the category? What is an example of an ordinal scale? What experiment supports this claim?
7. Interpretation How is a particular event interpreted or summarized? Does the graph show a main effect for "A"? What happened yesterday?
8. Feature specification What qualitative attributes does entity X have? What is the value of a qualitative variable? What is George like? What color is the dog?
9. Quantification What is the value of a quantitative variable? How much? How many? How many rooms are in the house? How much profit was made last year?
10. Causal antecedent What caused some event to occur? What state or event causally led to an event or state? How does warm air get to Ireland? Why is the kite going backwards?
11. Causal consequence What are the consequences of an event or state? What causally unfolds from an event or state? What happens to the warm winds when they reach the mountains? What are the consequences of double-digit inflation?
12. Goal orientation What are the motives behind an agent's actions? What goals inspired an agent to perform an action? Why did Roger move to Chicago? What was the purpose of the city's cutting taxes?
13. Enablement What object or resource anables an agent to perform an action? What device allows you to measure an earthquake? What do I need to bake this fish?
14. Instrumental/Procedural How does an agent accomplish a goal? What instrument or body part is used when an agent performs an action? What plan of action accomplishes an agent's goal? How does a person perform long division? How do you move a mouse on a computer?
15. Expectational Why did some expected event not occur? Why wasn't there a war in Iraq? Why doesn't this doll have a mouth?
16. Judgmental The questioner wants the answerer to judge an idea or to give advice on what to do. What do you think about the new taxes? What should I do to stop the fight?
17. Assertion The speaker expresses that he or she is missing some information. I don't understand what this message on the computer means. I need to know how to get to the Newark airport.
18. Request/Directive The speaker directly requests that the listener supply some information. Please tell me how to get a printout of this file.

Aschner, Gallagher, Perry, and Afsar (1961) identified the following five categories of thought processes:

1. Routine (R) questions. These refer to procedural matters, structure of class discussion, and approval or disapproval of ideas. Are there any questions?
2. Cognitive-Memory (C-M) questions. These require the use of recall or recognition in order to reproduce facts and other items of remembered content. What are the five steps in Knowles' self-directed learning model?
3. Convergent Thinking (CT) questions. The tightly structured framework of these questions requires the analysis or integration of given or remembered data, leading to one expected result. Based on this model, what are the goals of education?
4. Divergent Thinking (DT) questions. These questions permit an independent generation of ideas, directions, or perspectives in a data-poor situation. Why is learning necessary?
5. Evaluative Thinking (ET) questions. These questions are concerned with values rather than facts and convey a judgmental quality. Is this approach worth the effort? (pp. iv-viii)

As is evident from the variety of initiatives in relation to the taxonomy of questions, "question and answer" (Q&A) is an active area of research. It is powerfully stiumulated by the challenge of improving online search methodology. Some researchers have formulated vision and roadmap statements to guide future investigation:

• Jaime Carbonell, Donna Harman, et al. produced a Vision Statement to Guide Research in Question and Answering (Q-and-A) and Text Summarization (2000) which distinguished four "sophistication levels of questioner": "Casual "Template "Cub "Professional Questioner" Questioner" Reporter" Information Analyst"
• John Burger, Claire Cardie, et al. Issues, Tasks and Program Structures to Roadmap Research in Question and Answering (Q-and-A). (2002).

---

[Parts: First | Prev | Next | Last | All] [Links: To-K | From-K | From-Kx | Refs ]