This document discusses developing emotionally accessible MOOCs and affective learning analytics that can detect students' emotions, put them in context, and have the learning system respond accordingly. It provides examples of existing affective tutoring systems and their approaches. The document also presents ideas for a pilot study to examine how delivering course content with or without emotionally expressed voices impacts reading comprehension and attitudes.
Measures of Dispersion and Variability: Range, QD, AD and SD
The roadmap to emotionally accessible MOOCs
1. The roadmap to emotionally
accessible MOOCs
Submitted to L@S (rejected), AAATE2017 (short paper) and 3rd Annual UDL Symposium (UDL Spotlight)
Garron Hillaire, Francisco Iniesto, Bart Rienties
2. “Infants respond to how something is said
long before they understand what is said”
(Strohecker, 2004)
3. What is this exciting new affective learning
analytics coming to a learning system near you?!?!
“The Sytems he [D’Mello] and others are developing attempt to do
three things: detect physical markers, such as a smile or a furrowed
brow or dilated pupils, that indicate affect; put those signals into
context; and prompt the System to respond accordingly”
(Herold, 2016)
“Non-cognitive aspects include whether a student is frustrated,
confused, or distracted. More generally, students have mindsets (such
as seeing their brain as fixed or malleable), strategies (such as reflecting
on learning, seeking help and planning how to learn), and qualities of
engagement (such as tenacity) which deeply affect how they learn”
(Sharples et al., 2015)
Potential impact: medium Timescale: long (4+ years)
4. Case Study Situation
Selection
Situation
Modification
Attentional
Deployment
Cognitive Change Response
Modification
Affective Tutor Encouraging and
motivational
messages
Empathy and
Emotional
Displays
Gaze Tutor Content Repittion Attentional
Reorientation
message
UNC-ITSpoke Examination
Based Sub-
dialogs
ConfusionTutor Contradictory
Trilogs
Instructed
Reappraisal
Reappraisal
Affective
Learning
Companion
Affective Support
Messages
Non-verbel
Mirroring
(Blanchard, Baker, Ocumpaugh, & Brawner, 2013)
5.
6.
7. FutureLearn – Caring for Older People
NLTK Sentence Parse
1 Hello and welcome to Caring for older people: a partnership model.
2 Ageing and some of the challenges it represents can be a daunting experience for
older people, their family, their friends and healthcare teams alike.
3 The purpose of this course is to help you better understand and address these
challenges by:
understanding ageing from the perspective of older people and those who care for
them
approaching the care of older people within a collaborative and partnership-centred
model
promoting the wellbeing of older people by supporting their needs and preferences
creating a more positive environment for all by implementing a range of practical
strategies.
8. Spectral Analysis of Vocal Expression
Praat spectogram standard settings by Luucuma CC BY-SA 4.0
from https://commons.wikimedia.org
9. Rater-2
Positive Negative Neutral Mixed
Rater-1
Positive 7 0 0 0
Negative 0 7 0 0
Neutral 5 4 17 0
Mixed 0 0 0 0
Voice-2
Positive Negative Neutral Mixed
Voice-1
Positive 0 11 1 0
Negative 0 27 0 0
Neutral 0 0 0 0
Mixed 0 0 0 0
Positive Negative Neutral All
P R F P R F P R F P R F
Rater-1 0.00 0.00 0.00 0.18 1.00 0.30 1.00 0.04 0.07 0.68 0.20 0.10
Rater-2 0.00 0.00 0.00 0.26 0.91 0.40 0.00 0.00 0.00 0.07 0.25 0.11
Consensus 0.00 0.00 0.00 0.23 1.00 0.37 0.00 0.00 0.00 0.05 0.23 0.08
Positive Negative Neutral All
P R F P R F P R F P R F
Rater-1 0.33 0.57 0.42 0.19 0.71 0.29 0.00 0.00 0.00 0.09 0.23 0.13
Rater-2 0.42 0.42 0.42 0.26 0.64 0.37 0.00 0.00 0.00 0.20 0.31 0.23
Consensus 0.44 0.57 0.50 0.24 0.71 0.36 0.00 0.00 0.00 0.16 0.30 0.20
10.
11. Pilot - Ideas
• Interviews
• Do people have opinions about the voices they use and avoid using?
• Does emotion prediction about synthetic voice align with user perceptions?
• Do they have task specific voice preference?
• Reading Comprehension
• Incidental Emotion (how they feel walking in)
• Attitudes towards topic (do I hate math?)
• Variable on turning on and off emotion expression for positive and negative elements in the
content.
• Delivery without emotion instruction
• Positive delivered Positive
• Negative delivered Negative
• Positive delivered Positive and Negative delivered Negative
• Learning Task
• Reading comprehension evaluation (formative and summative)
13. References
• Blanchard, N., Baker, R., Ocumpaugh, J., & Brawner, K. (2013). I Feel Your Pain : A Selective Review of Affect-
Sensitive Instructional Strategies (pp. 1–16). Montreal, Canada.
• D’Mello, S., Jackson, T., Craig, S., Morgan, B., Chipman, P., White, H., … Graesser, A. (2008). AutoTutor
Detects and Responds to Learners Affective and Cognitive States. In Workshop on Emotional and Cognitive
Issues in ITS held in conjunction with the Ninth International Conference on Intelligent Tutoring Systems.
Montreal, Canada.
• Fitzgerald, E., Kucirkova, N., Jones, A., Cross, S., Ferguson, R., Herodotou, C., … Hillaire, G. (2017). Dimensions
of personalisation in technology-enhanced learning : A framework and implications for design, 0(0).
http://doi.org/10.1111/bjet.12534
• Gross, J. J., & Thompson, R. A. (2007). Emotion Regulation : Conceptual Foundations. In Handbook of
Emotion Regulation (pp. 3–24). New York: Guilford Press.
• Herold, B. (2016). Personalized Learning Based on Students’ Emotions : Emerging Research to Know.
Retrieved from
http://blogs.edweek.org/edweek/DigitalEducation/2016/01/personalized_learning_student_emotion_resea
rch.html
• Sharples, M., Adams, A., Alozie, N., Ferguson, R., FitzGerald, E., Gaved, M., … Yarnall, L. (2015). Innovating
Pedagogy 2015: Open University Innovation Report 4. Milton Keynes.
•