Search and Comprehension Processes in Learning from Text

1
Search and Comprehension Processes in Learning
from Text

• Cerdán, R., Vidal-Abarca, E., Gil, L., Gilabert,
  R., Martínez, T.
• University of Valencia

2
Comprehension Learning from Text

• Adjunct Questions Aids for Comprehension
  Learning

• Search tasks in complex documents ( Rouet
  Tricot, 1998)

• Evaluation (E) search goal strategy
• Selection (S) selection of information units
• Processing (P) extraction of relevant
  information
• Iterations of E-S-P
  cycles

• Pattern of search (Rouet, Vidal-Abarca,
  Bert-Erlboul Millogo, 2001)

• High level questions Review Integrate
• Low level questions Locate Memorize

3
Previous experiment to study Search
Comprehension processes (Vidal-Abarca, et al.,
2002)

• 22 University students
• 2 groups high vs. low level questions
• Task (on a computer screen)
• Reading long science text (1800 words)
• Searching info to Answer (HL vs. LL) questions
• Reading the question
• Re-reading the text (if neened)
• Writing the answer

Cycles
4

• Main results

Answering questions at a good level implied

• Reading questions fewer times.
• Selecting lower number of text segments
  (relevant non-relevant for the questions)
• Reading more relevant segments.
• Fewer answering cycles.

High level questions

• Reading questions more times and selecting more
  text segments (relevant non-relevant)

Low level questions

• Reading questions fewer times and selecting
  fewer text segments (relevant non-relevant)

5
Current experiment

• New Situation searching info to answer (HL vs
  LL) Qs, but NO prior reading text
• GOAL replicate prior results?
• Search comprehension processes in answering
  questions at good vs. poor level?
• Search patterns to answer high vs. low level
  questions?

6
Procedure

• 16 University students.
• 2 groups high level vs. low level questions
• Task on a computer screen (Read Answer)
• Searching info to Answer (HL vs. LL) questions
• Reading the question
• Reading the text
• Writing the answer

Cycles
7
ReadAnswer
1. Reading the question
2. Searching info to answer Q
8
Design
ANOVAs 2x2 Type of question x Comprehension
(Answering level)

• High L Q Integrating distant information
  many inferences
• Low L Q Locating especific information few or
  no inferences.

• Good comprehension (highest third score)
• Poor comprension (lowest third score)

9
On line measures
Evaluation Phase

• Number of times reading the questions.
• Time spent reading the questions.
• Word reading time per visit.

Selection Phase

• Number of total paragraphs visited.
• Number of relevant paragraphs.
• of relevant paragraphs.

Processing Phase

• of time reading relevant segments within each
  question.

Control Processes

• Number of QTW cycles Q (reading the question), T
  (reading the text), W (writing an answer),

10
Evaluation phase Times reading questions
Processing phase of time reading relevant
segments
Selection phase of relevant segments
Executive control processes number of QTW cycles
11
Summary

• Answering at a Good vs. Poor level

• Reading questions fewer times.
• Selecting a higher percentage of relevant
  segments ( especially in low level questions)
• Fewer answering cycles.

High vs.low level questions

• High level Reading questions more times,
  selecting more segments (relevant non-relevant)
  and using more QTW cycles.
• Low level Reading questions fewer times,
  selecting fewer text segments and using fewer QTW
  cycles.

12
Conclusions

• Good comprehenders effective search pattern
• Poor comprehenders loss in search task
• Pattern for High level questions
  ReviewIntegrate
• Pattern for Low level questions LocateMemorize