The principal design choices TANGISOFTwere 1. the use of direct-touch input,
2. the design of a keyboard as a tangible tool (allowing for two-handed interaction),
3. the use of a printed, rather than virtual keyboard layout, and
4. the augmentation of the physical paper with projected digital information.
4.4.1 Direct-touch input
The overhead of switching between tasks, in this case any task and a writing task, must be kept to a minimum. Observations of users performing tasks on traditional and digital tables have high- lighted both the fluent mix of activities and frequent rapid switching between tasks such as drawing
and writing (Tang, 1991; Bly, 1988; Scott et al., 2003). Such fluid transitions are hindered by the need to change between a direct-touch interaction and the use of a physical keyboard for text entry. Maintaining direct-touch input for text entry is an important component in the facilitation of fluid transitions between activities at the tabletop (Scott et al., 2003; Hinrichs et al., 2007).
4.4.2 The keyboard as a tangible tool
The use of soft keyboards in digital tabletop systems often requires users to change the location and orientation of the keyboard, in particular, foregrounding and backgrounding (or hiding) it when switching between tasks. Such move, rotate and display operations, when performed uni- manually, are executed sequentially thereby impose a significant additional load on the user. By contrast, bimanual interaction with objects allows for such task performance at a natural level of chunking, thereby leveraging significant cognitive advantages (Leganchuk et al., 1998; Hinckley et al., 1998; Buxton and Myers, 1986). Full utilization of such bimanual interaction can only be made possible by adding tangible (or graspable) qualities (Fitzmaurice et al., 1995; Terrenghi et al., 2007; Rekimoto et al., 2001).
4.4.3 Printed layout on paper
Instead of using a projected layout as in soft keyboards, TANGISOFTuses a printed keyboard and users press on the printed keys to input a character. Using the printed layout increases the tangible character of the device (as compared to simply having a physical handle on a virtual keyboard). The use of printed media allows very high resolution display of the characters, which addresses the characteristically low resolution of most existing tabletop projection systems.
4.4.4 Paper augmentation
Casting a keyboard as a tangible tool does not preclude the incorporation of additional digital properties. In particular, the flat paper layout allows the projection of digital information over the physical layout. Digital augmentation of the printed keyboard allows us to highlight the key that the stylus is moving over, and change the highlighting as that key is pressed. This provides important feedback to the user as to which key is actually being pressed and improves accuracy. The digital information augmented on the keyboard can even be in the form of highlighting for the predicted next possible characters which might be useful in some cases.
4.4.5 Comments on the design
The distinctive design feature of TANGISOFTis the hybrid design that gives tangible affordances (like two-handed manipulation) to virtual techniques (like soft-keyboards), yet maintains the direct- touch interaction. The conceptual design of TANGISOFTis layout independent. That is, although the QWERTY layout was used in the trials, due to its familiarity to most people, there is nothing
intrinsic to the design that prevents the use of other layouts, or even certain types of gesture based keyboards such as SHARK (Zhai and Kristensson, 2003). Therefore, issues of efficiency are not addressed here as efficiency is largely dependant on the used layout. In the case of a QWERTY keyboard layout, TANGISOFTbecomes as familiar as a traditional soft keyboard and as efficient.
4.5
Related work
Since the pioneering work of Wellner (1993) on the DigitalDesk, a wide range of design issues for tabletops have been explored and guidelines for designing collaborative systems on digital table- tops proposed (Scott et al., 2003; Morris, 2006; Wallace and Scott, 2008). Only a small number of projects have considered the problem of text entry for tabletops (Ryall et al., 2006; Wigdor et al., 2007). Indeed, the analysis of Hinrichs et al. (2007) is the only systematic account of the prob- lems and requirements of text entry on digital tabletops, and it establishes a criteria for evaluating existing approaches. The only text entry techniques designed specifically for digital tabletops are BubbleType (Hinrichs et al., 2008), the adaptive text input interface proposed by Hirche et al. (2008), and the SLAP keyboard (Weiss et al., 2009).
BubbleType is designed for a walk-up-and-use scenario in a public space. This means that people with diverse computer expertise and ages are expected to use the table, mostly for short periods of time. Such contextual factors require the text entry technique to be approachable (i.e. invites people to use it), as easy to use as possible, and even fun. The visual appearance of the technique and its integration within the overall application was also an important factor in the design. After an iterative design process that started with a circular layout, then a number of curved branches, the final design comprised two possible layouts, BubbleQWERTY and BubbleCircle (Figure 4.4). BubbleQWERTY, which followed the QWERTY layout, was found to be more comfortable to use than the circular layout and thus was the design adopted in the initial trials. Although designed for two-handed interaction, the keyboard could still be used with one finger or even in pen-based tabletop interfaces. The initial trials showed that people did not have difficulty using the keyboard as it was basically a fancy looking soft keyboard using the familiar QWERTY layout, but it also showed that most people typed on the keyboard using one finger rather than two hands. The design of BubbleType targeted a very specific context of use. Moreover, the design fixed the keyboard position and did not allow it to be moved or rotated. Thus, while the design process of BubbleType demonstrated good design practises following the guidelines stated by Hinrichs et al. (2007), BubbleType was not intended as a general solution to the problem of text entry for tabletops and it did not actually introduce any new interaction design concepts.
The approach to text entry proposed by Hirche et al. (2008) was designed specifically for two- handed text entry on multi-touch multi-finger tables. The technique relies on a small number of buttons with prediction and hints to minimize finger movement and draws on aspects of a 10-key phone keypad. The main motivations were to overcome the problem of locating keys on the flat
Figure 4.4: The two layouts used in BubbleTYPE. BubbleQWERTY (left) and BubbleCircle (right) (Hinrichs et al., 2008)
and featureless surface of the tabletop and to reduce the visual attention required during typing. The proposed keyboard adapts its position, size, and orientation based on the initial placement of the fingers when the keyboard is activated. No formal results for performance and ease of learning have been reported, but due to its novelty, a certain level of practice is required. The technique specifically targets mutli-touch multi-finger tables and is not appropriate for single touch and pen- based tabletops.
Weiss et al.’s (2009) SLAP widgets, which allow a tangible cast of transparent silicon to be augmented with digital information, shares some of the tangible-virtual design qualities of TANGISOFT. A SLAP keyboard (Figure 4.6) is in fact one of the proposed widgets introduced with the aim of maintaining the tactile nature of physical keyboard yet allow for digital informa- tion to be projected on the transparent keys. As part of a widget set, however, this work does not provide any detailed analysis as to the usability of the SLAP keyboard. Like the adaptive keyboard, the SLAP keyboard was designed exclusively for multi-touch interaction.
Figure 4.6: The SLAP keyboard (Weiss et al., 2009)
Approaches to text entry, and direct touch in particular, in settings other than tabletops have been thoroughly studied, including the use of optimized keyboard layouts (MacKenzie and Zhang, 1999), improving text entry with prediction (Magnien et al., 2004), gestural alphabets (Goldberg and Richardson, 1993; Blickenstorfer, 1995), and gesture based techniques (Zhai and Kristensson, 2003; Perlin, 1998; Ward et al., 2000). The hybrid design of TANGISOFT can still be used with many of these direct-touch techniques to enhance them with tangible characteristics.