Content uploaded by Jochen Rick
Author content
All content in this area was uploaded by Jochen Rick
Content may be subject to copyright.
From DigiQuilt to DigiTile:
Adapting Educational Technology to a Multi-Touch Table
Jochen Rick and Yvonne Rogers
The Open University
Department of Maths, Computing, and Technology
Walton Hall, Milton Keynes, MK7 6AA, UK
{j.rick, y.rogers}@open.ac.uk
Abstract
To realize the potential of multi-touch tables, interaction
designers need to create meaningful applications for them
in real-world contexts. One convenient shortcut towards
that end is adapting a meaningful application from another
interface paradigm. In this paper, we detail the process of
adapting DigiQuilt, a single-user desktop educational tech-
nology, to DigiTile, a collaborative multi-touch application.
With this case study, we concretely demonstrate the utility of
adapting and how previous research and theory can inform
that process. In particular, we show how learning theory (1)
motivated the transition from the desktop to the multi-touch
table, (2) guided the design process, and (3) informed the
evaluation.
1 Introduction
Each new interface requires studies of how the technol-
ogy can best be appropriated. Early work tends to focus
on technical innovations and basic usability. “Demo apps”
that demonstrate the features and potential of the new UI
are created. Increasingly, more sophisticated applications
are developed to solve real world problems in ecologically-
valid settings.
Interaction designers are tasked with creating these ap-
plications. Do they start from scratch? Or, can they adapt
existing applications, developed for other user interfaces, to
the new hardware? Here, we make the case for the latter.
We detail our own success in adapting DigiQuilt, a single-
user desktop application, into DigiTile, a multi-user multi-
touch tabletop application.1
1We chose to change the name for the multi-touch version for two rea-
sons. First, it makes it easier to distinguish between the two in writing.
Second, quilting is seen differently in the U.S., where DigiQuilt was con-
ceived, and the U.K., where DigiTile is used.
2 Adapting Applications
Adapting is one of the best and most natural ways we
have to understand novel interfaces. A novel interface is a
new medium. As with any new medium, it is difficult to
predict its affordances. One of the best practices for un-
derstanding a new medium is remediation—the practice of
taking techniques and practices that work in one domain
and applying them to the new domain [1]. Many new me-
dia are created and explored in this matter. For instance,
early sound recording was based on a parallel to written
text; Edison envisioned his phonograph as writing sounds
[4]. By applying the practices of the better understood
medium, a better understanding of the new medium can be
achieved. With experience and innovation, new uses par-
ticularly suited to the new medium emerge. So, while Edi-
son envisioned a machine that could both read and write
sounds, recorded sound technology came into prominence
with the application of pre-recorded musical records. Based
on these principles, a particularly fruitful way to investigate
new user interfaces is to apply ideas from already existing
applications to the new interface.
Adapting an application also has several practical advan-
tages. First, the plethora of existing applications provide
a good source of inspiration for design. Even if only one
percent of all desktop applications would make for viable
multi-touch table applications, there would still be plenty
of worthy candidates, given the large amount of desktop ap-
plications. Much of good design is appropriating previous
good ideas [11]. A successful application embodies several
working ideas. Adaptation allows the designer to build on
these.
Second, adapting can shorten development time by lever-
aging on the previous design work. A speedy development
cycle can be critical to research on novel interfaces. It can
be difficult for application-based research to keep up with
new hardware innovations; by the time an application has
been built for specific hardware, it is out of date. There
is only a limited amount of time when research results can
have an important impact. If the novel interface becomes
commercially viable, work on the novel interface can move
out of the research domain before significant results can be
achieved. Adapting an existing application can cut that de-
velopment time down significantly, because the designer of
the new application can appropriate the design work put into
the existing application.
Third, the existing application and the adapted applica-
tion can inform each other. For novel user interfaces, a com-
parison to a similar application for an established interface
can provide researchers a point to compare and contrast.
Thus, adapting can also aide in evaluating the new interface.
While adapting has its advantages, there are also chal-
lenges that come with adapting to a new interface. Depend-
ing on the nature of the application, overcoming these chal-
lenges can range from easy to difficult. For the easy case,
consider using a tabletop for baggage screening. In baggage
screening, the user is trying to identify dangerous objects,
such as guns, from an x-ray image. Having multiple users
view the image from different angles can help increase ac-
curacy [3]. To adapt this task to a multi-touch table is trivial:
The display just needs to be placed horizontally. This works
well, because the interaction (clearing the bag or signaling
for a hand search) does not change much (if at all). For ap-
plications where the graphical user interface does not need
to change, middle-ware that maps touch input into desktop
commands, such as mouse movements, can speed up the
adaption process [28].
Where some applications are easy to adapt, others re-
quire serious design and development work. While reme-
diation works to a certain extent, the remediated practice
is usually different to the original practice [1]. Changing
the interfaces changes the activity. For these adaptations,
guiding theory of the application domain and previous re-
search results can inform the process. First, these sources
can provide motivation for adapting a specific application.
Second, they can inform the concrete design. Third, they
can help develop criteria for evaluation. In this paper, we
concretely demonstrate how learning theory and previous
research informed these three phases of adapting DigiQuilt
to a multi-touch table.
3 From DigiQuilt to DigiTile: A Case Study
DigiTile was created for the ShareIT project,2which
aims to further the understanding of novel interfaces for
supporting collaboration in real-world situations. One path
for supporting collaboration with computers is single dis-
play groupware [26], where multiple people can simulta-
neously interact with a shared display. This user paradigm
2http://shareitproject.org
is being supported by a number of new technologies. One
such technology is multi-touch tabletop computers. Table-
top computers have a wide design space, especially when
part of a larger ubiquitous computing infrastructure [19].
Collaboration tables [23], tabletop computers that allow
multiple people to interact, are still a fairly novel user in-
terface. Hardware innovations are still being published on a
regular basis [6, 7, for example]. Yet, a reasonable hardware
basis exists for conducting HCI research. For example,
MERL’s DiamondTouch [2] has been actively researched
since 2001. Software toolkits, such as DiamondSpin [24]
and reacTIVision [9], help developers overcome some of
the more vexing problems of developing software for these
platforms. With this hardware and software basis, research
on tabletop computing can focus on creating applications.
Yet, designing applications from scratch can be a difficult
and time-consuming task.
To shortcut this process, we adapted an existing applica-
tion, DigiQuilt, to study the potential of multi-touch table-
tops to support collaboration in an authentic setting. In par-
ticular, DigiTile was designed to support learning, a task
where collaboration can often be beneficial. Designing
computer-based learning environments is an established re-
search field with established criteria [25] for success. While
multi-touch table technology is relatively new, there have
also been published research on users using multi-touch
tabletops for learning [14, 16, for example]. Finally, there
is the previous research on DigiQuilt. DigiTile drew on all
of these areas.
This case study first introduces DigiQuilt as an interest-
ing educational technology. Next, it motivates adapting Di-
giQuilt to a multi-touch table. Then, it details the signifi-
cant design decision in creating DigiTile and the rationale
behind those decisions. Finally, the plans for evaluating Di-
giTile are discussed.
3.1 DigiQuilt
DigiQuilt is a construction kit [17] for learning about
math and art by designing patchwork quilt blocks [13]. It
is based on the the instructional theory of constructionism,
which holds that people learn particularly well when de-
signing personally-meaningfully public artifacts [15]. In
constructionism, learners are motivated to construct the ar-
tifact, because it is meaningful to themselves and because
it can be shared with others. In DigiQuilt, third and fourth
graders assemble colored pieces into a square quilt block
(Figure 1).
Clicking a color button in the palette (1), turns the pieces
at (2) that color. Clicking on a piece creates a copy of
that piece, which can be dragged into the quilt block (3)
or into the work area (4). In the work area, pieces can
be rotated and assembled into more complex pieces. The
1
2
3
4
5 6
7
Figure 1. Using DigiQuilt to create a quilt
block design (actual 3rd grader’s design)
complex pieces can then be copied and transferred into the
quilt block. The software keeps a complete history of the
quilt block, which learners can navigate (6) to revert un-
wanted changes. Once finished, the design can be saved (5).
Saved designs can be opened and edited. Designs can be
printed out to be displayed in the classroom, used as wrap-
ping paper, etc. In user studies, children enjoyed creating
designs with DigiQuilt and proudly showed them to their
classmates, teacher, and family [12].
In addition to being aesthetically pleasing, the patch
work quilts also lend themselves to mathematical analysis.
The designs embody fraction concepts and are often sym-
metric. For instance, the design in Figure 1 is half red and
half yellow; it is also diagonally symmetric. To highlight
the learning goals, learners are given increasingly difficult
challenges to accomplish, such as creating a design that is
half red or creating a design that is horizontally symmet-
ric. To help learners reflect on fractions, the fraction of the
whole that a certain color covers is displayed next to that
color (Figure 1, Point 1). To help learners reflect on sym-
metry, different grids (Figure 1, Point 7) can be overlaid on
the quilt. These tools help learners reflect on the targeted
concepts.
3.2 Motivation
The previous section introduced DigiQuilt as a success-
ful and compelling single-user desktop application; how-
ever, that by itself is not enough to justify adapting it to a
multiple-user tabletop application. Adapting to a new in-
terface requires examining both the application and at the
implementation level. Could this application benefit signif-
icantly from the move to the new interface? If not, why
adapt. How hard is it to implement the transition to the new
interface? Even if the application would be compelling, im-
plementation issues might make it impractical.
3.2.1 Application
The primary reason to adapt DigiQuilt to a multi-touch ta-
ble is the potential to support collaborative learning. Col-
laboration is an established method for enhancing learning.
Single display groupware has been shown to better sup-
port collocated collaborative learning than taking turns on
single-user application [8].
One relevant theory of collaborative learning is
Roschelle’s [21] theory of convergent conceptual change—
when two learners work together with a reflective tool (a
tool that responds to user input to reflect the embedded do-
main concepts), they tend to converge on an understand-
ing that is better than either would achieve independently.
When learners work together on a challenge, they naturally
articulate how they would solve the challenge, based on
their understanding. If their strategies clash, it triggers them
to justify their understanding. The reflective tool allows the
learners to demonstrate or test their understanding. Thus, an
understanding can be confirmed or rejected. As they work
on the challenge together, learners’ conceptual understand-
ings do not just converge with each other, but also with the
domain concepts embodied in the tool.
DigiQuilt is a reflective tool, giving feedback when
changes are made (e.g., the fraction updates when a new
piece is placed). So, we expect that two learners working
with DigiQuilt to engage in convergent conceptual change.
There is also practical evidence that collaboration might be
warranted: DigiQuilt users often make comments on oth-
ers’ designs and sometimes want to take over the controls
to show something [12].
Roschelle’s model of collaboration is particularly com-
pelling, because it shows how collaboration between equals
can benefit both partners. On the flip side, if collaborators
are not evenly skilled, as is often the case in a classroom
setting, a more appropriate model of collaboration may be
that of Vygotsky’s [29], where a more skilled person helps
a less skilled person gain competence. No matter which
model of collaboration (even or uneven) is appropriate to
a specific pair, we would still expect the collaboration to
benefit learning.
3.2.2 Implementation
While implementation is a challenge when actually adapt-
ing, it is worth considering beforehand. If there is not a
good fit between the source application and the new inter-
face, it may be difficult to successfully adapt. The interface
may prove awkward or the application may lose its com-
pelling nature in the process.
DigiQuilt’s existing interface is a good match for a multi-
touch table. It is a fairly simple application with a small
number of user actions. The primary action is to move
pieces to the work area and quilt block. While this is easy
enough with a mouse, simply dragging the shapes with a
finger on a multi-touch table could be an even better phys-
ical mapping. Early prototypes of DigiQuilt were carried
out with physical shapes on paper. While this physical ap-
proach had significant drawbacks,3moving pieces by hand
was an intuitive interface [12]. Besides dragging pieces, the
main interactions in DigiQuilt are pushing buttons, some-
thing that translates directly to a multi-touch table. Digi-
Quilt does make use of the keyboard to name designs. This
small amount of typing can be achieved with a virtual on-
screen keyboard (see Figure 4).
3.3 Design: To DigiTile
As outlined above, the basic interactions of DigiQuilt are
well supported by a multi-touch table, but there are still im-
portant design decisions to be made in adapting. In this sec-
tion, we detail how learning theory and previous research
helped us negotiate these design challenges. We organize
the subsections around design features well documented in
the literature on multi-touch tables—group / table size, ori-
entation, and collaboration.
3.3.1 Group / Table Size
A first problem to consider is group size and table size [22].
Based on the theory of convergent conceptual change, two
is the ideal number of users: Two can challenge and help
each other without spending undue amount of time coordi-
nating.
We wanted the learners to collaborate on one design, so
the table needed to be small enough to allow intended users
(10–12 year-olds) to both reach every part of a central tile.
The smaller DiamondTouch table (32 inch diagonal) eas-
ily allowed for these young users to reach almost all of the
table.
3While a physical interface was intuitive, it failed in other ways. First,
it was hard to properly constrain the physical world: Users could easily
overlap tangible pieces in ways that did not allow for fractional analysis.
Second, physical pieces do not allow for several useful operations easily
implemented in software: assembling simple pieces into a complex piece
(such as an L shape), creating copies of these complex pieces, swapping
colors, reverting to an earlier design point, etc. Third, with a large number
of colors and shapes, a physical implementation would require an over-
whelming amount of different pieces.
3.3.2 Orientation
Unlike a desktop display, different users need not have the
same perspective on a tabletop display. They could sit
across from each other, around the corner, or next to each
other. This makes it difficult to maintain the desktop orien-
tation, where eyes up equals the top of the screen and eyes
down means the bottom of the screen [27]. Developing soft-
ware without a dominant orientation can be tricky [24].
For DigiTile, different positions would change how
learners view the tile. If seated around a corner, horizontal
symmetry for one partner would be vertical symmetry for
the other. If seated across from each other, the tile would ap-
pear upside down. For more abstract designs, this would not
be a problem; however, many DigiQuilt designs are based
on real-life objects which have a preferred orientation (e.g.,
a stick figure). While adults tend to favor working across
from each other, children often prefer working next to each
other [23]. So, it made sense to position the learners next
to each other to share the same orientation. This configura-
tion (dyads next to each other) allows DigiTile to maintain
a desktop-style orientation (Figure 2).
3.3.3 Software Implementation
DigiTile is implemented in Squeak [5], a multimedia-
enabled cross-platform open-source Smalltalk. A small
Java application reads DiamondTouch events and broad-
casts them to Squeak via OSC (Open Sound Control).
Squeak then handles the events. While it was not designed
for creating applications for a multi-touch table, it was rel-
atively easy to adapt Squeak for this purpose. For instance,
all Squeak interfaces can be rotated.
More pressing for DigiTile, Squeak allows for multiple
concurrent mouse pointers, which can be a challenge for
some UI toolkits [23]. Mapping touch input to mouse in-
put is not entirely straightforward, since hovering and click-
ing do not work quite the same as for a mouse. In our
mapping, any touch was treated as a mouse down condi-
tion; further movements were treated as mouse drag move-
ments. This simple mapping matched well with DigiQuilt’s
interface, since its main interaction techniques were drag-
ging tile pieces and pressing buttons. Other common mouse
functions, such as pressing secondary buttons and scrolling,
were not necessary.
DigiQuilt itself was implemented in Squeak. While the
source code for DigiQuilt was available, we choose to im-
plement DigiTile from the ground up. Like most software
which has been slowly evolved over an extended time by
several developers, DigiQuilt’s code base had serious prob-
lems. The size of the pieces was hard coded. The model
and view were tightly coupled. The file format was opti-
mized for size, not flexibility. We took this opportunity to
rewrite the application with a cleaner design, being espe-
1
2 3
4
5
6
Figure 2. Using DigiTile to create a tile
cially careful to decouple model and view; thus, different
views for multi-touch application and the desktop applica-
tion can share the same underlying model.
3.3.4 Collaboration
As with DigiQuilt, the main focus of DigiTile (Figure 2) is
the quilt block (1). Since the learners are positioned next
to each other, we placed a palette area on both the left (2)
and right (3) side. Practically, each learner has their own
palette to use. This positioning allows learners to feel like
they have specific ownership over part of the interface. To
encourage sharing, we created a shared work area (4). We
intentionally used an odd number of work zones, so that one
work zone would be in the middle, implying no left or right
ownership. Like in DigiQuilt, the palette buttons are labeled
with the name of the color. In classroom use of DigiQuilt,
the labelling proved to be useful in allowing children to talk
about their designs [12]. Similarly, we believe that these
labels can help aide dialogue (e.g., students can easily dif-
ferentiate between lime and green).
After some user testing, it was obvious that dropping a
piece accidentally was fairly common in the multi-touch
case. In addition, partners often disapproved of a specific
change. For these reasons, navigating the history was more
important for DigiTile than DigiQuilt. We developed a
graphical history (5) that allowed users to easily back up
to a previous design. In contrast, DigiQuilt only provided
undo and redo buttons (Figure 1, Point 6).
While Squeak supports concurrency at the implementa-
tion level, concurrency at the application level is not always
a good idea. For instance, saving an application that is con-
currently being worked on is problematically ambiguous.
To solve this problem, we implemented a menu bar (6).
Figure 3. Selecting a grid to test for symmetry
Figure 4. Saving a tile design
When a menu is pressed, the rest of the interface becomes
unavailable, signified graphically by darkening the applica-
tion area (Figure 3). Other elements of the interface were
relatively insensitive to concurrent interaction. Even if two
tiles were dropped simultaneously, they could be processed
separately. The only practical difference being that two his-
tory positions would have been created, rather than one.
Using the menu, users launch file functions (new, open,
save, etc.) and can apply different grids. Different inter-
faces appear for saving (Figure 4) and opening (Figure 5)
designs. The grid menu enables feedback on symmetry.
When a line of symmetry is selected, the corresponding icon
appears next to the “Grid” menu label; that icon shows a star
when that line of symmetry is valid (see the diagonal icon
in Figure 3).
Because of collaboration, the learning process in Digi-
Figure 5. Opening an existing tile
Tile is different than the learning process in DigiQuilt. To
effect convergent conceptual change, DigiTile use should
encourage different perspectives for different users. To fur-
ther this, we provide different ways for the two learners to
analyze the quilt. DigiQuilt displays the fraction that is
covered by a color on the respective colored button. Di-
giTile allows the user to change the representation from re-
duced fraction to least-common-divisor fraction to percent-
age to visual pie chart. One learner could display his or her
palette with percentages, while the other displays reduced
fractions. This allows for us to create challenges like “a tile
that is 1
2red and 50% yellow.” This challenge allows the
learners to discover that 1
2and 50% are the same. Using
multiple-linked representations to represent a mathematical
concept leads to deeper math understanding [10]. This strat-
egy would not work as well in the single-user DigiQuilt.
3.4 Evaluating DigiTile
Since the focus of this paper is the design process, we do
not have room to detail study results. Instead, we show how
learning theory and previous work informed the design of
the evaluation. Previous research on DigiQuilt has already
engaged many of the questions of whether tiling is a useful
medium to engage with mathematics and art. This allows
us to concentrate on questions that arise from adapting the
interface to a multi-touch table. As HCI researchers, we are
interested in how the multi-touch table interface can (better)
support collaborative authoring.
An initial user study was conducted with 10–12 year
olds to investigate whether and how pairs of learners col-
laborate using DigiTile. In particular, we were interested in
whether the kind of reflective dialogue, that Roschelle based
convergent conceptual change on, occurred in this design
task. After a brief familiarization phase, we assigned pairs
Figure 6. Working together on a difficult chal-
lenge
mathematical challenges. The goal was to see how pairs
would collaborate when faced with a difficult challenge—
one that required significant thought and work. Some chal-
lenges proved trivial and were accomplished in less than
a minute with minimum planning and negotiation between
the participants. We increased the difficulty until the task
became challenging. At that point, the pairs had a harder
time forming and trialing approaches. Because both par-
ticipants could interact simultaneously, it was necessary for
one participant to recruit the other in trialing an approach
(Figure 6).
This initial study confirmed that DigiTile could encour-
age the kind of collaboration we were looking for. It also
gave us feedback on how learners dealt with the different
tools (graphical history, fraction representations, grid menu,
etc.) in the full version of DigiTile. Even before this study,
we believed a tailored version, leaving out features not nec-
essary to the challenge at hand, would suit the learning task
better. Yet, we left all features in as the study was largely
formative, allowing us to see how intended users engaged
the tools and how the interface could be improved.
In a follow-up study, we worked with DigiTile in a class-
room, where the teacher felt that DigiTile’s focus on frac-
tions would be appropriate. We worked with the teacher and
an educational specialist to streamline DigiTile for that set-
ting. To reduce complexity, we reduced the number of col-
ors and shapes (Figure 7). Reducing the shapes meant that
the fractions would no longer be as complex. With all the
shapes available, it was fairly easy to place pieces to create
denominators of 256 or 400 (larger than the learners were
familiar with); in the simplified version, denominators are
unlikely to reach above 64 (with a 4x4 grid) or 100 (with a
Figure 7. Using the simplified version for a
training task
5x5 grid). Reducing the number of colors removed colors,
such as cyan, that children were not that familiar with. It
also gave us more space to represent the fractions. As one
of the learning goals in the curriculum was converting from
fraction to decimal representation, we showed both the frac-
tion and decimal representation simultaneously (underneath
the color name).
As researchers, we were still interested in how different
features of the interface would affect the collaboration. One
way to enforce collaboration is to allocated resources un-
evenly. So, we created another version of DigiTile that split
the colors between the users (Figure 8). We also designed
challenges, such as design a tile that is 1
2red and 1
2yel-
low, that take advantage of this split. In this second study,
we will compare the collaboration of the simplified version
with that of the split version. Is the split effective in pro-
moting collaboration? Is there more reflective dialogue in
one case? As of the writing, this study has been conducted,
but the data has not been analyzed.
4 Conclusions
Creating meaningful applications for novel interfaces is
an important challenge for HCI researchers. There are dif-
ferent approaches to solve this problem. For instance, the
Equator Project applied creative exploration as the basis for
development [20]. Here, we made the argument for adapt-
ing an existing application. With the DigiTile case study,
we concretely demonstrated both the benefits and the chal-
lenges in implementing this approach.
While DigiTile is only one example of adapting to a
multi-touch table, we believe many of the issues presented
here are universal. Implementation issues of group / table
Figure 8. Using the split-palette version for a
training task
size, orientation, and collaboration are inherent in the de-
sign space of multi-touch tables. By giving a concrete ex-
ample, we also demonstrate the specific benefits of adapt-
ing. For example, while code reuse may seem like a tempt-
ing benefit of adapting, we show the difficulties in realizing
this. In contrast, design work and research on the source
application was beneficial in motivation, design, and evalu-
ation. DigiQuilt had an extensive design process, including
paper prototypes and classroom-evaluated design iterations
[18]. It took several years to evolve DigiQuilt; it took less
than half a year to create DigiTile.
Adding collaboration changed how the application was
used. Rather than seeing this as a problem, we see this as
an opportunity to utilize the strengths of the new interface.
Multi-touch tables are better suited to collocated collabora-
tion than a desktop setup. As learning can benefit from col-
laboration, DigiTile is more than just a port of DigiQuilt.
The learning scenario and the interface were changed to
take advantage of the multi-touch interface. The represen-
tation of fractions and the role of the challenges were sig-
nificantly changed to suit the new learning paradigm. By
adapting DigiQuilt, we were able to create a novel appli-
cation to support collaborative learning with a multi-touch
table.
5 Acknowledgements
The work is part of the ShareIT project funded by the
EPSRC, grant number EP/F017324/1. We would like to
thank our other collaborators in that project for their help-
ful feedback: (alphabetically) Sheep Dalton, William Farr,
Rowanne Fleck, Amanda Harris, Eva Hornecker, Paul Mar-
shall, Richard Morris, Nadia Pantidi, and Nicola Yuill. Spe-
cial thanks go to K.K. Lamberty for supporting us in adapt-
ing DigiQuilt and Kim Bryant for help in creating the fig-
ures. We thank MERL for loaning us the DiamondTouch.
References
[1] J. D. Bolter and R. Grusin. Remediation: Understanding
New Media. The MIT Press, Cambridge, MA, 1999.
[2] P. Dietz and D. Leigh. DiamondTouch: A multi-user touch
technology. In Proceedings of UIST ’01, pages 219–226,
New York, 2001. ACM Press.
[3] C. Forlines, C. Shen, D. Wigdor, and R. Balakrishnan. Ex-
ploring the effects of group size and display configuration
on visual search. In Proceedings of CSCW ’06, pages 11–
20, New York, 2006. ACM Press.
[4] L. Gitelman. Scripts, Grooves, and Writing Machines: Rep-
resenting Technology in the Edison Era. Stanford University
Press, Stanford, CA, 1999.
[5] M. Guzdial and K. Rose, editors. Squeak: Open Personal
Computing and Multimedia. Prentice Hall, Upper Saddle
River, NJ, 2002.
[6] J. Y. Han. Low-cost multi-touch sensing through frustrated
total internal reflection. In Proceedings of UIST ’05, pages
115–118, New York, 2005. ACM Press.
[7] S. Hodges, S. Izadi, A. Butler, A. Rrustemi, and B. Bux-
ton. ThinSight: Versatile multi-touch sensing for thin form-
factor displays. In Proceedings of UIST ’07, pages 259–268,
New York, 2007. ACM Press.
[8] K. M. Inkpen, W. ling Ho-Ching, O. Kuederle, S. D. Scott,
and G. B. D. Shoemaker. "This is fun! We’re all best friends
and we’re all playing": Supporting children’s synchronous
collaboration. In Proceedings of CSCL ’99, pages 252–259,
Mahwah, NJ, 1999. Lawrence Erlbaum Associates.
[9] M. Kaltenbrunner and R. Bencina. reactivision: A
computer-vision framework for table-based tangible inter-
action. In Proceedings of TEI ’07, pages 69–74, New York,
2007. ACM Press.
[10] J. Kaput. Linking representations in the symbolic systems
of algebra. In S. Wagner and C. Kieran, editors, Research
agenda for mathematics education: Research issues in the
learning and teaching of algebra, pages 167–194. National
Council of Teachers of Mathematics, Reston, VA, 1989.
[11] J. Kolodner. Case-Based Reasoning. Morgan Kaufmann,
1993.
[12] K. K. Lamberty. Getting and Keeping Children Engaged
with a Constructionist Design Tool for Craft and Math. PhD
thesis, Georgia Institute of Technology, Atlanta, GA, May
2007.
[13] K. K. Lamberty and J. L. Kolodner. Exploring digital quilt
design using manipulatives as a math learning tool. In
P. Bell, R. Stevens, and T. Satwicz, editors, Proceedings of
ICLS 2002, pages 552–553, Mahwah, NJ, 2002. Lawrence
Erlbaum Associates.
[14] M. R. Morris, A. M. Piper, A. Cassanego, and T. Wino-
grad. Supporting cooperative language learning: Issues in
interface design for an interactive table. Research Report 8,
Department of Computer Science, Stanford University, Palo
Alto, CA, 2005.
[15] S. Papert. Situating constructionism. In I. Harel and S. Pa-
pert, editors, Constructionism, pages 1–13. Ablex, 1991.
[16] A. M. Piper, E. O’Brien, M. R. Morris, and T. Winograd.
SIDES: A cooperative tabletop computer game for social
skills development. In Proceedings of CSCW ’06, pages 1–
10, New York, 2006. ACM Press.
[17] M. Resnick, A. Bruckman, and F. Martin. Pianos not
stereos: Creating computational construction kits. Interac-
tions, 3(6):41–50, 1996.
[18] J. Rick and K. K. Lamberty. Medium-based design: Extend-
ing a medium to create and exploratory learning environ-
ment. Interactive Learning Environments, 13(3):179–212,
2005.
[19] Y. Rogers, Y.-K. Lim, and W. R. Hazlewood. Extending
tabletops to support flexible collaborative interactions. In
Proceedings of TABLETOP ’06, pages 71–78, Washington,
DC, 2006. IEEE Computer Society.
[20] Y. Rogers, M. Scaife, E. Harris, T. Phelps, S. Price,
H. Smith, H. Muller, C. Randell, A. Moss, I. Taylor, D. Stan-
ton, C. O’Malley, G. Corke, and S. Gabrielli. Things aren’t
what they seem to be: innovation through technology inspi-
ration. In Proceedings of DIS 2002, New York, 2002. ACM
Press.
[21] J. Roschelle. Learning by collaborating: Convergent con-
ceptual change. In T. Koschman, editor, CSCL: Theory
and Practice of an Emerging Paradigm, pages 209–248.
Lawrence Erlbaum Associates, Mahwah, NJ, 1996.
[22] K. Ryall, C. Forlines, C. Shen, and M. R. Morris. Explor-
ing the effects of group size and table size on interactions
with tabletop shared-display groupware. In Proceedings of
CSCW ’04, New York, 2004. ACM Press.
[23] S. D. Scott, K. D. Grant, and R. L. Mandryk. System guide-
lines for co-located, collaborative work on a tabletop dis-
play. In K. Kuutti, E. H. Karsten, G. Fitzpatrick, P. Dour-
ish, and K. Schmidt, editors, Proceedings of ECSCW 2003,
pages 159–178, Netherlands, 2003. Kluwer.
[24] C. Shen, F. D. Vernier, C. Forlines, and M. Ringel. Dia-
mondSpin: an extensible toolkit for around-the-table inter-
action. In Proceedings of CHI ’04, pages 167–174, New
York, 2004. ACM Press.
[25] E. Soloway, M. Guzdial, and K. E. Hay. Learner-centered
design: The challenge for HCI in the 21st century. Interac-
tions, 1(2):36–48, April 1994.
[26] J. Stewart, B. B. Bederson, and A. Druin. Single display
groupware: A model for co-present collaboration. In Pro-
ceedings of CHI ’99, pages 286–293, New York, 1999. ACM
Press.
[27] A. Tang, M. Tory, B. Po, P. Neumann, and S. Carpendale.
Collaborative coupling over tabletop displays. In Proceed-
ings of CHI 2006, New York, 2006. ACM Press.
[28] E. Tse, S. Greenberg, and C. Shen. GSI Demo: Multiuser
gesture / speech interaction over digital tables by wrapping
single user applications. In Proceedings of ICMI ’06, pages
76–83, New York, 2006. ACM Press.
[29] L. S. Vygotsky. Mind in Society: The Development of
Higher Psychological Processes. Harvard University Press,
Cambridge, MA, 1978.