Multilingualism and Justintime Feedback on the Web

1
Multilingualism and Just-in-time Feedbackon the
Web

• Leveling the learning curve for a truly
  international virtual experience(changing the
  tools instead of forcing the culture)

Radu Luchianov HOTLab, 03 October 2003,
Carleton University, Ottawa
2
Interactive examples
Problem examples
Linguistic vandalism Clickeaza vs.
Apasa Switched word order specific
modifiers Icon opacity save/open icon
evolution The WorldWideWait Guess-the-function

• Baloon help
• Monade (under development) (graph editor)
• eastcree.org
• read-along
• stories database
• MoStaCon

Tools

• Flash
• DHTML
• Java
• Other scripting options with cross-platform
  support

3
Research objectives

• Design Implement theory in functional prototypes
• Test Verify/Validate theoretical and practical
  assumptions (e.g.
• multilingualism vs. internationalization
• JIT feedback vs. decoupled or serial help
  sources)

Existing technology
Theory (MonDoc)
Design
Experiment
4
Some ubiquitous feedback loops
Supports everything else
Currently dissociated,spatially and
culturally,due to economic constraints
Reply-interaction
End-user
Maintenance
Collaboration
Function
Designer
Mutualcomprehension
Tester
Form
Programmer
Expression
Userside
Continuous specialization
Systemside
Existing Web technologiesallow for JIT
interactivity. JITF multilingual presentation is
still problematic.
5
Layers of interaction

• Representation based on abstraction (from sensory
  perception)
• Features follow function
• Function is context-based

• Unexpected effects
• Cross-layer feature interactions
• due to label scarcity or conceptual proximity
  (e.g. Search/Browse, Browse/Select)
• Memory effects on action recall

6
Design Functionality

• In real products, function is based on applied
  physics (engineering, ergonomics)
• In software, function is based on correct
  instructions (exhaustively verified algorithms
  and interfaces)
• but the computer is not the only element at
  work it is not enough to provide functionality,
  the system is also responsible to make that
  functionality accessible
• fit the interaction model the user expects, or
  JITF
• guide the user to understand the internal model
  of the product, or
• change the internal model to fit specific user
  cultures

Badly designed stroller, requires a lot of force
to handle, thus breaks fast and stalls very often.
7
Affordance

• Features of an object which suggest its possible
  uses
• Context-based reasoning is heavily influenced by
  perceived affordances.
• However, affordance, especially for symbols is
  mostly conventional, thus culture-dependent.

Move the cursoron the screen with the mouse
Undo
8
Aesthetics

• used to be neglected, though it positively
  influences perceived affordance and helps focus
  attention

Test
Test
Link
Plain Button
Nice, realistic Button
9
Some stumbling blocks

• for Multilingualism
• Different grammatical rules (creating
  multilingual generators is usually overkill)
  (computer-oriented NLP research vs. pragmatic
  interaction)
• Uncharted cultural differences (inter- and
  intra-group differences in symbol assignment and
  norms)
• Inaccessibility to functional and formal
  differences in designers vs. end-users (non-HCI
  designers create tools based on their own
  cultural biases differences are apparent only
  after product is distributed)
• Current technological hiatus (Unicode)
• for Just-in-time feedback
• Reducing perceptual load (hiding complexity) can
  increase user confusion (especially in less
  technically-inclined cultures)
• Layout modification issues
• General reluctance to Virtual media
• Economic issues (income from training)

10
Thank You