Title: THE INFLUENCE OF SCREEN READERS ON WEB COGNITION
1THE INFLUENCE OF SCREEN READERS ON WEB COGNITION
- Tony Stockman Oussama Metatla Queen Mary,
University of London
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
2OVERVIEW
- Examine implications of SR technology for web
cognition - Report findings from survey and study of
collaborative web use - Propose draft taxonomy of errors in collaborative
web interaction - Examine potential role of non-speech sound in
addressing identified problems
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
3FEATURES OF SR WEB INTERACTION
- Default linear model of page presentation
- No ambient representation of location on page
- No representation of spatial layout
- No immediate indication of information density
- Little to assist formulation of mental model of
page
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
4CURRENT FUNCTIONALITY
- Focus Jaws, WE and VO.
- Cursor key navigation strength and weakness of
analogy with other apps - Other mechanisms
- By listing of navigating forward/back between
links, headers, frames, tables, forms, text
elements, markers etc. - Forms mode
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
5OVERVIEWS
- Important but often under supported/neglected
- Typically lists no. of links, frames, headers,
and forms with reminder of related hot keys
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
6CURRENT NON-SPEECH SOUND
- Jaws schemes
- Window-eyes events
- VoiceOver defaults
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
7VoiceOver
- Embedded in OS
- Factors in switching
- Group v. DOM mode navigation
- Overview followed by interaction model
- Non-speech sound more visible
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
8IMPLICATIONS FOR COGNITION AND INTERACTION
- Hindered by linear navigation EG pageing through
search results - Hot keys and markers help, but ignore density of
info and spatial layout - consequences for
collaboration - Overviews neglect spatial layout, ordering and
esthetics, - Tables are navigable but lack overviews
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
9DESCRIBING WEB PAGES
- Employed widely known pages
- Characteristics of sighted descriptions
- Relatively short but covering main features
- Column layouts, colours, mood, style, pictures,
emotional response to message - Characteristics of VI descriptions
- Longer, more factual and granular, more focus on
function and usability - Conclusion the two groups approach web tasks
from widely differing contexts
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
10COLLABORATIVE TASKS
- Involved pairs of sighted and visually impaired
users - Both could read the web pages used
- One gave instructions while the other performed
the task - The tasks involved
- Simple information searches
- Comparisons of data values
- Navigating pages and filling forms
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
11VISUALLY IMPAIRED INSTRUCTOR
- These tasks were performed generally quite
straightforwardly - VI user was generally familiar with sites
- Sighted users perspective generally compensated
for difference in views of each user - Sources of problems
- Screen-readers focus unavailable to sighted user
- Sighted user referring to spatial layout
unavailable to screen-reader
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
12SIGHTED INSTRUCTOR
- Substantially more problems
- Point in task unclear because screen-reader focus
unavailable to sighted user - Sighted references to spatial layout
- CAPTCHEs
- Non-standard form controls
- Column headers not spoken on forms
- Dynamic updating of form fields
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
13TOWARDS A TAXONOMY OF COLLABORATIVE ERROR 1
- Location disconnects
- Layout disconnects
- Missing objects
- Navigation disconnects
- Contextual disconnects
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
14TOWARDS A TAXONOMY 2
- Affordance disconnects
- Modal disconnects
- Hollistic disconnects
- Multi-focus disconnects
- Esthetics disconnects
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
15NON-SPEECH AUDIO
- Audio is inexpensive and widely used
- Screen-readers only gradually adopting limited
non-speech sound, eg forms and progress bars - Growing body of knowledge on how to design and
use (www.ICAD.org) - Range of techniques that could be examined as
part or whole solutions to problems described
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
16NON-SPEECH AUDIO 1
- Monitoring for dynamic changes
- The structure of Earcons could for EG reflect
object type and nature of update - Ambient sound might convey esthetics and/or
interaction mode - Auditory icons might signal affordance
open/closed - Spatial sound might convey overall layout,
density, locations of users
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
17NON-SPEECH AUDIO 2
- Example of non-speech auditory overviews compared
with speech, more like a glance - Spearcons for typical radio button options
- Crucial to avoid auditory overload, masking etc.
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc
18CONCLUSIONS
- Speech-only model struggles to convey rich web
content and enable increasingly complex
interactions - This is highlighted by gap in first impressions
of common web pages - Cross modal web collaboration os subject to a
range of disconnects due to differences in
presentation and interaction - Non-speech audio is an under-used mechanism that,
with careful design, could help to address some
of the issues sited
Interaction Media Communication, Department of
Computer Science, Queen Mary University of
London http//www.dcs.qmul.ac.uk/research/imc