Do Bystanders and Dialog Participants Differ in Preferences for Telecommunications Channels? -- The...
-
Upload
juliette-harbach -
Category
Documents
-
view
213 -
download
0
Transcript of Do Bystanders and Dialog Participants Differ in Preferences for Telecommunications Channels? -- The...
Do Bystanders and Dialog Participants Differ in Do Bystanders and Dialog Participants Differ in Preferences for Telecommunications Channels?Preferences for Telecommunications Channels?
-- The Effects of Noise and Delay ---- The Effects of Noise and Delay --
Nigel WardNigel WardAnais G. RiveraAnais G. RiveraAlejandro VegaAlejandro Vega
University of Texas at El PasoUniversity of Texas at El Paso
Why?
The MysteryThe Mystery
Mobile telephone conversations are often banned
because they can be annoying to bystanders.
But why are they more annoying than face-to-face conversations?
Is it the volume? Perhaps in part, but cell phone conversations are more annoying even when no louder than face-to-face conversations (Monk et al. 2004a)
Is it the lack of an audible interlocutor, inducing a psychological “need to listen”? Perhaps in part, but this doesn’t explain the annoyance (Monk et al. 2004b)
Is it the Channel?Is it the Channel?
Transmission Rating Factor (ITU-T Rec G.107)
R = Ro – Is –Id –Ie-eff + A
But what about bystander preferences?
Channel properties affect user perceptions.The E-model can predict these, for infrastructure design purposes.
Ro = signal-to-noise ratio Is = simultaneous impairment Id = delay impairment factor Ie-eff = equipment impairment factor (e.g. codec) A = advantage factor
Potential SignificancePotential Significance
Hypothesis 1:
For telecommunication channels, bystanders preferences differ from users preferences
If true, there may be a technological fix to the problem
Today: In a Possible Future:
NO B>70PHONES!
Perceptions of DelayPerceptions of Delay
We know that delay affects talkers’ perceptions
(Emling & Mitchell 1964)
How Line Delay Affects How Line Delay Affects Conversation DynamicsConversation Dynamics
Likely 1st Order Effects: • more awkward silences • more overlaps
Likely 2nd Order Effects: • more explicit turn-taking cues
annoyance
lack of audibleinterlocutor
feeling ofembarrassment
channelproperties
involuntarylistening
• delay• noise• echo
negative attitudes tocell phones
handsetproperties
• lack of sidetone• low volume
negativeimpressionsof talker
• bossy• show-off• insensitive• etc.
changedspeaking style
• loud• exaggerated prosody• etc.
differentsituationat remote end
• incongruous speaking styles• incongruous topic• lack of shared awareness
cognitiveeffects
• uncertainty about receipt• frustration• cognitive load
Likely Effects on BystandersLikely Effects on Bystanders
Hypothesis 2Hypothesis 2
Hypothesis:
Bystanders dislike channel delay more than do talkers
where we measure “more” relative to a standard impairment: codec quality
Experiment DesignExperiment Design
High NoiseLow Delay(Cn)
Bystanders’Perception
Talkers’Perception
goodgood
Low NoiseHigh Delay(Cd)
less goodgood
T Δ = TCn - TCd B Δ = BCn - BCd
Hypothesis 2: compared to talkers, bystanders dislike delay more i.e. T Δ < B Δ, i.e. T Δ - B Δ < 0
unfortunately not supported
by Wilcoxon sign test, chi-square, or matched-pairs t-test
G.711350 ms
GSM-FR150 ms
Software/Hardware ConfigurationSoftware/Hardware Configuration
• channels emulated on Linux machines
• talkers in different rooms
extra delay (CD) or extra noise (CN) recorder
ProceduresProcedures
1. welcome
2. dialog with Cn or Cd
3. questionnaire
4. dialog with Cd or Cn
5. questionnaire
6. debrief
1. welcome
2. overhear
3. questionnaire
4. overhear
5. questionnaire
6. debrief
Two Talkers Two to Eight Bystanders
usually with same stimuli, different judgessometimes with same judges, different stimuli
(when talkers were later used as bystanders)sometimes with same judges, same stimuli
(when talkers later listened to recordings of themselves)
Experiment Conditions (1)Experiment Conditions (1)
Distance from Talker to Bystanders• > 4 meters• ~ 2 meters• ~ 0.5 meters
Distractors• pizza and friends• magazines• none (paying attention)
Dialog Content Cn Cd• multi-digit number exchange• free dialog• single-digit number exchange
Experiment Conditions (2)Experiment Conditions (2)
Presentation• live• recorded, played over speakers• matched-content extracts, headphones
Subjects• naive students• experts
Survey Format• forced choice• 4 choices• 11 point scales
ResultsResults
TΔ = talker preference re channel quality (Cn – Cd)BΔ = bystander preference re less-annoying (Cn – Cd)
ResultsResultsOn the last experiment:
Subjects’ preferences for Cn over Cd,
as talkers and as bystanders
SummarySummarySummary results for Hypothesis 2:• Across 59 dialog stimulus-pairs, in various conditions - bystanders seemed to dislike Cn more than did talkers,
contrary to hypothesis 2- however the difference was small and not consistent
(averaging 1.42 vs 1.47 on a scale from 0 to 3)• Even under unrealistically exaggerated conditions, line delay does not consistently impact bystanders
Summary Results for Hypothesis 1: • No evidence that bystanders and dialog participants differ
in preferences
The Mystery Remains
Do Bystanders and Dialog Do Bystanders and Dialog Participants Differ in Preferences Participants Differ in Preferences
for Telecommunications Channels?for Telecommunications Channels?
-- The Effects of Noise and Delay ---- The Effects of Noise and Delay --
Nigel WardNigel WardAnais G. RiveraAnais G. RiveraAlejandro VegaAlejandro Vega
University of Texas at El PasoUniversity of Texas at El Paso
Do Bystanders and Dialog Participants Differ in Do Bystanders and Dialog Participants Differ in Preferences for Telecommunications Channels?Preferences for Telecommunications Channels?
-- The Effects of Noise and Delay ---- The Effects of Noise and Delay --
Nigel WardNigel WardAnais G. RiveraAnais G. RiveraAlejandro VegaAlejandro Vega
University of Texas at El PasoUniversity of Texas at El Paso
Why?
Dialog-Based Evaluation of Mobile Phone Infrastructure
Phase 1
A. Your opinion of the connection you have just been using. (Please place a line crossing the axis at the appropriate point.)
excellent
good
fair
poor
bad
10
8
6
4
2
0
firstdialog
seconddialog
B. What differences did you notice between the two connections?
C. What do you think affected your ratings of the two connections? date ___________session ________subject A Brecording# 1 _________recording# 2 _________
T
excellent
good
fair
poor
bad
10
8
6
4
2
0
Dialog-Based Evaluation of Mobile Phone Infrastructure
Phase 2
A. Sometimes conversations can be annoying to bystanders, independent of the content, due to the way the the speaker was talking. Considering the potential for annoyance due to the speaking style, please give your opinion of the sample. (Please place a line crossing the axis at the appropriate point.)
excellent
good
fair
poor
bad
10
8
6
4
2
0
firstdialog
seconddialog
B. What differences did you notice between the two samples?
C. What do you think affected your ratings of the two samples? date ___________session ______subject A Brecording# 1 _________recording# 2 _________
B
excellent
good
fair
poor
bad
10
8
6
4
2
0
Dialog-Based Evaluation of Mobile Phone Infrastructure
Phase 3
date ___________session ______subject A Brecording# 1 _________recording# 2 _________
R
A. Sometimes conversations can be annoying to bystanders, independent of the content, due to the way the the speaker was talking. Considering the potential for annoyance due to the speaking style, please give your opinion of the sample. (Please place a line crossing the axis at the appropriate point.)
excellent
good
fair
poor
bad
10
8
6
4
2
0
firstdialog
seconddialog
B. What differences did you notice between the two samples?
C. What do you think affected your ratings of the two samples?
excellent
good
fair
poor
bad
10
8
6
4
2
0