Music and SelfGenerated Images: Applying Dual Coding Theory, Generative Theory of Reading Comprehens

1
Music and Self-Generated Images Applying Dual
Coding Theory, Generative Theory of Reading
Comprehension, and Cognitive Load Theory to
Web-Based Music Instruction George J.
HenikEducational Communication and Technology
ProgramNew York UniversityUnited
Stateswww.georgehenik.com

• ed.ss.webmaster_at_nyu.edu
• georgehenik_at_yahoo.com

2
Problem

• What is an effective computer/web-based
  theoretical approach for teaching music?

3
Proposed Solutions

• Dual-Coding Theory (DCT) (Clark Paivio,
  1991 Paivio, 1971, 1990)
• Sound and Image Connection (from DCT)
• (Clark et al, 1974 Datteri, 2000 Henik,
  2002 Lipscomb and
• Kendall, 1994 Hiraoka and Umemoto,
  1981)
• Generative Theory of Reading Comp.
• (Wittrock, 1990)
• Visual Complexity Cog. Load
• (Henik Allen, 2002 Miller, 1956 Sweller,
  1994)

4
The Study

• Image and Music Connection?
• Cant assume music to language mapping
• Limited to interval recognition tasks
• Distance between two notes
• Short
• Simplest music ear-training task
• User-Generated Images
• Users generate meaning

5
The Study

• Must take interface into consideration
• Would the keyboard be viewed as one piece or many
  separate parts?

6
Research Question 1

• What is the relationship between using
  self-generated images for interval recognition
  and
• pre-test scores
• music background
• drawing background
• attitudes towards the study

7
Research Question 2

• What are the effects of a more visually complex
  interface on learning musical intervals through
  user-generated images?

8
Sample

• 2 groups of Graduate Students
• 17 from Pilot 1
• 14 from Pilot 2

9
Materials

• Computers connected online
• Flash 5 plugin
• Headphones

10
Procedures

• Pretest
• Music Questionnaire
• Treatment
• Practice Screen
• Three separate intervals to learn (M3, P4,
  Tritone)
• Learner draws a picture to each
• Pilot 1 Pilot 2

11
Procedures
12
Procedures

• Drawing Questionnaire
• Posttest
• 35 - 45 Minutes

13
Results (RQ1)

• What is the relationship between user-generated
  images for interval recognition and
• a) prior music knowledge
• b) prior drawing knowledge
• c) music interest and
• d) drawing interest?

14
Results (RQ1)

• Sig. correlations on the exit survey
• Felt confused (CONFUSED)
• -.683, p.007 posttest
• Able to hear differences (HEARDIF)
• .551, p .041 pretest
• Became bored (BORED)
• -.604, p .022 posttest

15
Results (RQ1)

• Hierarchical multiple regression
• Model 1 Pretest
• Model 2 Drawing and Music Background
• Model 3 CONFUSED, HEARDIF, BORED
• R2 .872 (p .011)
• POSTTEST 8.497 1.485PRETEST -
  0.711MUSIC_BACKGROUND 0.477DRAWING_BACKGROUND
  0.515CONFUSED 1.113HEARDIF 0.599BORED
  ? (MUSIC_BACKGROUND, CONFUSED, and BORED were not
  significant in the model).

16
Results (Pilot 1 vs 2)

• What are the effects of a more visually complex
  interface on learning musical intervals through
  user-generated images?

17
Results (Pilot 1 vs 2)

• ANCOVA
• Need to determine group differences between Pilot
  1 2, want to control for the pretest.
• Pretest scores as the covariate
• Treatment group as a fixed factor

18
Results (Pilot 1 vs 2)

• ANCOVA
• Covariate (pretest) significant
• (F 7.385, p .011)
• IV (treatment group) decreased but not
  significantly.
• Overall R2 .254

19
Conclusions

• Why was music background found not to be
  significant in Pilot 2?
• Particular sample had a lower range for music
  background than for either Pilot 1 or drawing
  background

20
Conclusions

• Only question from Exit survey that was
  significant in the overall model was the
  participants ability to hear differences
  (HEARDIF)
• Though the other questions had a correlation to
  either pre or post tests, this relationship is
  not singularly important when mixed with other
  factors

21
Conclusions

• ANCOVA showed that the pretests were significant
  but no significant differences between groups
• Addition of keyboard combined with low music
  background may have increased cognitive load

22
Conclusions

• By generating visual information in conjunction
  with aural information, users are undergoing a
  deeper level of cognitive processing - and
  learning these intervals
• How effective is this method?
• Further research

23
Implications

• Only one of many future studies to help develop
  guidelines for musical multimedia environments.
• Careful attention to interface

24
Implications

• Take advantage of multimedia!
• Power over traditional learning through images
  which can be dynamically displayed
  (user-generated)

25
Future Research

• Using homogenous groups of music students
• Comparing different levels of images (none,
  static, user-generated) in a repeated measures
  design
• Effect of learner differences on interval
  recognition tasks (eg., field dependency)

26
Thank You

• Any Questions?
• http//www.georgehenik.com