Head Mounted Displays and Deaf Children: Facilitating

Head Mounted Displays and Deaf Children: Facilitating
Sign Language in Challenging Learning Environments
Michael Jones
M. Jeannette Lawler
Eric Hintz
Dept. of Computer Science
Brigham Young U.
Provo, UT
[email protected]
Dept. of Physics and
Astronomy
Brigham Young U.
Provo, UT
Dept. of Physics and
Astronomy
Brigham Young U.
Provo, UT
Nathan Bench
Fred Mangrubang
Mallory Trullender
Dept. of Computer Science
Brigham Young U.
Provo, UT
Dept. of Education
Gallaudet U.
Washington D.C.
Mantua Elementary School
Fairfax, VA
ABSTRACT
There is much we do not know about how deaf and hard of
hearing children learn and there is even less we know about
how deaf children experience sign language in HMDs. The
driving principle behind our work is to deliver instruction
in sign language rather than in written captions. This is
particularly important for young children who are learning
sign language as their first language.
Compared to spoken language acquisition by children who
hear, a child who is deaf or hard-of-hearing often experiences
significant acquisition delays with their first language [5]. It
is estimated that 95% of school-age deaf and hard of hearing
children are born to hearing parents [6]. These children often
do not begin learning sign language until entering school and
may only receive fluent language input during school hours.
Improved sign language learning may lead to increased
learning in a second language such as English. Research
shows “that children who learn through their first (minority) language for as long as possible not only tend to have
improved final achievement, but also their English language
skills tend to develop to a higher level than those who were
taught through their second language with some first language support” [4].
In this paper we explore the configuration of and potential benefits of head-mounted displays for education done
in sign language in difficult environments. Difficulties arise
when students cannot see or are not looking directly at the
signer. Our purpose is to evaluate the comfort and utility
of viewing sign language in an HMD as perceived by a child
who communicates primarily in sign language through the
use of ASL video presented through an HMD. This may enable both teachers and deaf children to interact in new ways
by allowing students to view instruction wherever they may
look.
Head-mounted displays (HMDs) are evaluated as a tool to
facilitate student-teacher interaction in sign language. Deaf
or hard-of-hearing children who communicate in sign language receive all instruction visually. In normal deaf educational settings the child must split visual attention between
signed narration and visual aids. Settings in which visual
aids are distributed over a large visual area are particularly
difficult. Sign language displayed in HMDs may allow a deaf
child to keep the signed narration in sight, even when not
looking directly at the person signing. Children from the
community who communicate primarily in American Sign
Language (ASL) participated in two phases of a study designed to evaluate the comfort and utility of viewing ASL in
an HMD.
Categories and Subject Descriptors
H.5.2 [Information Interfaces and Presentation]: User
Interfaces; H.1.2 [Information Systems]: User/Machine
Systems
1.
INTRODUCTION
Spoken and signed languages are each difficult to use in
certain environments. For example, spoken languages are
difficult to use in noisy environments, but headphones with
speakers and directional microphones allow people to communicate more easily. Signed languages are visual rather
than auditory, and they are difficult to use in a different set
of environments than spoken languages. However, as with
spoken languages, technology can facilitate communication
in these settings.
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than
ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from [email protected]
IDC’14, June 17–20, 2014, Aarhus, Denmark.
Copyright 2014 ACM 978-1-4503-2272-0/14/06 ...$15.00.
http://dx.doi.org/10.1145/2593968.2610481 .
2.
RELATED WORK
Methods have been developed for delivering written English captions to deaf or hard-of-hearing students in the
classroom (such as C-Print [8]). While these methods are effective for children with strong English (or any other written
language) reading skills, they are not effective for children
with poor reading skills. Over 30 years of educational test-
317
ing in the United States, the average reading level is below
the fourth grade level (which included children ages 9-10 in
the United States). It included hearing children in grades
K through 12. The fourth grade level is normalized against
the reading level of their hearing peers. [7].
Several groups have explored methods for deliving captions in a planetarium setting. A planetarium is a particularly difficult environment to present ASL. One of the early,
well documented efforts is reported by DeGraff and Hamil
[2]. DeGraff and Hamil used a slide projector to project
captions near the horizon of the planetarium dome. Daniel
[1] details a slightly different way of using captions in a live
show. The words are either projected near the object of interest on the dome or a green arrow directs student’s eyes
to the relevant area of the dome.
In Grice [3] we find the beginnings of a more modern approach to captioning systems. This is a major move into
devices for planetariums and theaters. The first device they
tested were Virtual Vision glasses that showed captions over
the right eye in a small screen. Grice reported that most
people put these away after a few minutes and experienced
a “dizzying effect.” Grice experimented with an LED display
system that was mounted behind the audience. This system
would display the captions in reversed text. For the demo,
Plexiglas was mounted on specific seats that would then reflect the captions back to the viewer correctly. Finally, Grice
tested a Vacuum Fluorescent Display (VFD). This is a box
that attached to a seat in front of the individual and had
captions run across the screen. The planetarium installed
four VFD captioning systems that could each support three
people.
It was apparent during our review that very little work
has been done using HMDs in Deaf education. Most work
focused on using captions to address the issue of relaying
information to deaf participants. However, this approach
does not cater to younger deaf children as their reading skills
have not fully developed.
3.
Figure 1: The three HDMs used in evaluations. The
displays are all monocular and include fully occlusive
on the left, partially occlusive in the middle, and
see-through with a half-silvered mirror on the right.
Display
Virt. Realities VR1
Vusix Tac-eye
Laster PMD-G3
Table 1:
fov
Res.
40◦ 800x600
29.5◦ 800x600
50◦ 800x600
Color
24 bits
8 bits
24 bits
Weight
3 oz.
1.81 oz.
3 oz.
guage but who are not part of the investigation team were
recruited to interact with and interview participants in order
to avoid bias.
A video record was kept for all interactions with the subjects. Cameras were positioned to allow us to view sign
language used by both the interviewers and the participants
during interviews and focus groups. Cameras were also used
to record interactions between the subject and the HMD
hardware. The videos were later translated into English and
coded both for verbal content and subject actions by both
ASL-speaking and English-speaking investigators.
In each evaluation the subject could adjust the size, position, and brightness of the ASL signer video. Initially,
participants adjusted the video using a laptop keyboard, but
this required looking at the keyboard, which proved distracting while trying to watch ASL and a video at the same time.
We later modified the system to use a video game controller
rather than the laptop keyboard to collect the positioning
data. All adjustments were logged for later analysis.
METHODS AND RESULTS
We conducted a two-phase evaluation of comfort and utility for HMDs used to convey ASL to children in logistically
challenging informal education environments. A total of 18
deaf or hard-of-hearing students who communicate primarily
in sign language participated in this portion of the study. In
the first phase, 8 participants provided subjective feedback
after watching a short astronomy based video on a screen
with ASL narration viewed through an HMD. We provided
software for repositioning the ASL video in the display and
recorded changes made by each subject. In the second phase,
10 participants watched a 20-minute planetarium show with
the narration provided in ASL. Five watched the narration
projected directly onto the planetarium dome and 5 watched
the narration in an HMD.
Test subjects were drawn from a local deaf school and a
summer university program for high school students. In each
case we limit the study to children who are deaf or hard-ofhearing and who communicate primarily in ASL. All of the
participants were between 13-18 years of age.
Because the participants in this study communicate in
ASL, care was taken to minimize linguistic barriers. Interactions with the participants were direct ASL-to-ASL communications without any intervening interpreter. Deaf and
hard-of-hearing individuals who use ASL as their native lan-
3.1
Displays
The three displays used in our evaluations are shown for
comparison in Figure 1. We used only monocular displays
to maximize the amount of light reaching the eye and because adding stereoscopic 3D to the display does not necessarily improve comprehension when viewing sign language.
The display on the left of Figure 1 is a fully occlusive Virtual Realities VR1. The center display is a Vusix Tac-eye
partially occlusive display. The display on the right is a
Laster PMD-G2 see-through display with a half-silvered mirror that blocks some incoming light. Table 1 contains the
diagonal viewing angle, resolution, color depth, and weight
for each display. Viewing sign language at a resolution of
800x600 is not likely to negatively impact comprehension.
Weaver et al.’s study [9] of ASL comprehension and video
resolution found that novice signers could observe and reproduce specific signs with equal success when learning those
signs from video rendered at 640x480, 320x240, or 160x120
on a mobile phone screen.
318
The second phase was conducted in the planetarium so
we could spread visual information over an entire viewable
hemisphere in a controlled environment. Ten participants
ages 13-15 were recruited from a local school for the Deaf.
Spreading visual information out over the entire planetarium
dome creates a logistically challenging environment for ASL
instruction. The duration of the planetarium show allowed
us to observe the use of the displays for comfort and utility
over a longer period of time. In this phase we showed a video
called “New Horizons” which was produced specifically for
projection onto the dome of a planetarium. We obtained the
English transcript of the narration for the show and asked a
Certified Deaf Interpreter (CDI) to translate the narration
into ASL.
3.2.1
The primary codes found among all participants are split
focus, fit and position, signer position, occlusion, and attention. Among all groups, position and fit was the most
common theme with split focus as the second most common.
The position and fit theme includes comments related to the
position and fit of the HMD itself on the participants’ head
or face. The split focus theme contains comments related
to splitting visual focus between the signer in the HMD and
the external world. Each theme is addressed below.
Signer Position. We provided two ways for subjects to
change the position of the signer. Subjects could move the
signer in the display and subjects could move the display
itself. The majority of the participants had the HMD positioned on the top-right of the right lens of the glasses with
one exception: One participant being left-eye dominant preferred the top to middle-left of the left lens.
Once the HMD was properly positioned on the glasses,
subjects exhibited a slight preference to moving the sign
language presenter down and toward to the center of the
subjects’ field of view. This preference was evident in both
the subjects’ comments and in the positioning log. Some
subjects turned or tilted their heads in order to place the
signer at the center of their field of view. Subjects may have
turned their heads to reposition the signer because they did
not know how to adjust either the HMD itself or the position
of the signer in the video.
Split attention. Subjects talked about the challenge of
splitting their visual focus between the signer in the HMD
and either the video or the people around them. All of
the subjects mentioned splitting visual focus between the
signer and the screen. Two made positive comments which
are given in the subsequent section. Comments from the
other six, three male and three female, indicated difficulties
focusing on both the interpreter in the HMD and the movie
projected onto the screen. A male subject reported, “I feel
like it is separate, and it is jarring to look back and forth.”
In contrast, two female participants stated that they liked
being able to see both the signer and the visual presentation
at the same time. One said, “I liked the HMD; it was good.
I liked being able to see the screen and the interpreter–to see
the speech and the sign interpreter on the screens. I liked
that. It was neat.”
HMD Fit. Subjects described discomfort related to the
fit of the HMD. These issues are concerned primarily with
occlusion and with issues related to weight and balance. The
Vuzix Tac-eye is designed for military and tactical use where
ruggedness and durability are more important than weight
Figure 2: Example of what was seen in an HMD by
a participant. A) Shows the ASL interpreter in the
HMD view. B) Scene from the video watched on a
screen. C) Illustration of a child wearing an HMD.
All three displays listed in Table 1 were too large, bulky,
and heavy for use by children. The Laster offered the largest
viewing angle but had a small eye box. The eye box is the
volume of space in which the display is correctly aligned with
the eye to allow viewing of the full image. A few participants
spent quite a bit of time adjusting the display to make sure
that the full field of view was visible.
3.2
Themes from First Evaluation
Evaluation
The first phase of evaluation was conducted using video
shown on a flat screen in a room. Four female and four
male participants, ages 15 through 18 were recruited from
a university summer camp for youth who are deaf. Each of
the subjects watched a five-minute video while the narration
in ASL was delivered through an HMD (Figure 2).
The individual subjects were interviewed about their experience. Eight children from the first phase were brought
together into two mixed-gender groups of four to participate
in a focus group. In both the individual and the focus group
interviews participants were asked to discuss their opinions
both of the concept and its execution. We encouraged them
to tell us when, where, and how they thought an HMD could
be used. We also asked about issues of design and comfort.
We coded both the transcripts and the video recordings
of the interviews and the focus groups. Open coding of
the transcripts allows us to identify themes from intentionally open-ended questions. We defined codes for the video
recordings that related to comfort and utility. Video recordings of interviews and focus group discussions were translated from ASL to English by a deaf interpreter. All transcripts and video recordings were coded by three members
of the research team.
319
sual inputs with minimal eye movement thereby minimizing
the effort and latency involved with shifting visual attention between the signer and what is being presented in the
environment.
Based on the feedback we received and from our observations, a smaller and lighter HMD designed for the geometry
of a child’s head should be developed. In addition, the use
of participatory design will be paramount to maximize the
potential of a child specific HMD design. The outlook for
the successful use of HMDs for deaf children in educational
settings is promising and should be explored.
Table 2: Recorded final adjustment of signer position in an HMD.
Subject
1
2
3
4
6
7
8
9
Eye
right
left
right
right
right
right
right
right
Horizontal
- 2.25
5.625
1.125
0
0
-1.6875
-2.25
-1.125
Vertical
0
0
2.8125
0
0.5625
0
-3.375
0
5.
and balance. Adding some weight to the frame of the glasses
may have corrected the balance problem but would have
added more weight to the HMD unit. In a comment typical
of others, a female subject said, “It felt uneven having to
compensate for my head being pulled to the side.”
We observed some participants, that used the Laster display, continued to make adjustments with their hands to
steady the display in proper viewing position. This may be
due to the small eye box found on the Laster display.
3.2.2
Adjustments of Signer Position
Subjects could use software controls to reposition and resize video of the ASL signer. All adjustments were recorded
for later analysis and a summary of repositioning data is
shown in Table 2. The data suggests a bias toward viewing
the signer in the center of the field of view. In the table, negative numbers represent movement to the viewers’ left and
positive numbers represent movement to the viewers’ right.
For example, movement in the negative direction by a a
subject viewing ASL with the right eye indicates movement
toward the center of the subject’s field of view. Adjustments
by subjects 1,2,7,8 and 9 moved the signer toward the center
of the field of view while adjustments by subject 3 moved the
signer away from the center. Vertical adjustments were less
common. The data in Table 2 only includes changes made
in software. Subjects also repositioned the signer in their
field of view by tilting their heads and physically moving
the display on their face.
4.
REFERENCES
[1] L. Daniel. Planetarium for the deaf. Planetarian, 3(1),
1974.
[2] J. V. DeGraff and F. Hamil. Seeing stars. Planetarian,
1(2), 1972.
[3] N. Grice. Resources for making astronomy more
accessible for blind and visually impaired students.
Astronomy Education Review, 5:154, 2006.
[4] P. Knight and R. Swanwick. Working with Deaf Pupils:
Sign Bilingual Policy into Practice. Routledge, 2002.
[5] M. Marschark and P. Hauser. How Deaf Children Learn:
What Parents and Teachers Need to Know. Perspectives
on Deafness. Oxford University Press, USA, 2011.
[6] R. E. Mitchell and M. A. Karchmer. Chasing the
mythical ten percent: Parental hearing status of deaf
and hard of hearing students in the united states. Sign
Language Studies, 4:138–163, 2004.
[7] S. Qi and R. E. Mitchell. Large-scale academic
achievement testing of deaf and hard-of-hearing
students: Past, present, and future. Journal of Deaf
Studies and Deaf Education, 2011.
[8] Rochester Institute of Technology. C-print computer
software, 2014. Accessed January 2014.
[9] K. A. Weaver, T. Starner, and H. Hamilton. An
evaluation of video intelligibility for novice american
sign language learners on a mobile device. In
Proceedings of the 12th international ACM
SIGACCESS conference on Computers and
accessibility, pages 107–114. ACM, 2010.
CONCLUSIONS
We identified two sources of discomfort in using HMDs
to view ASL by children who are deaf or hard-of-hearing
and who communicate primarily in ASL. First, the displays
we used were too large and bulky for children to use effectively. Second, participants struggled to split their attention between the signer in the HMD and the external world.
This may be due to the design of HMDs tested, their utility,
and the novelty of the device. In addition to these limiting factors, finding the appropriate number of deaf children
from our community was an issue. Future recruitment from
other Deaf communities outside our area will be necessary
to broaden the pool of potential participants.
We consistently found a slight preference to position the
signer in the center of the field of view but this was not
a universal preference. This was demonstrated both in the
ASL position data and the physical adjustment of the HMD
on the participants head. Placing the signer in the center of
the field of view may allow the subject to switch between vi-
320