In order to provide alternative computer input, we designed an isometric, tongue operated device: Tonguepoint. The design rationale and a preliminary experiment are presented in this technical note. Results show that, after 30 minutes practice and adjustment, the subjects could use the Tonguepoint at a performance level that was only 5-50% slower than finger isometric pointing. Further improvements are expected.
Disability, tongue, mouth, isometric device, input device, alternative access.
The success of the modern WIMP (Windows,. Icons, Menus and Pointers) interfaces have made pointing devices, such as the mouse, the Trackball, and in keyboard isometric joysticks (TrackpointTM), an essential part of computer systems. The majority of the current input devices are designed for hand use. Due to various physical disabilities, however, there is an unfortunate population of users who are unable to use hand operated input methods . Hand pointing devices are also problematic for restrictive environments and tasks that need both hands to be completely dedicated to a specific operation. Such tasks might include driving, piloting, or underwater exploration. As a result, alternative input devices, such as the foot mouse, the foot trackball, head mounted pointers, chin mounted joysticks, eye tracking, and voice recognition have all been developed for interaction with WIMP interfaces. Although all of these alternative input devices are somewhat successful, they are also limited in many aspects.
The somatosensory and motor cortex homunculi (See  for an illustration) show that the tongue and the mouth occupy an amount of sensory and motor cortex that rivals that of the fingers and the hand. Furthermore, unlike the eyes, which have rich cortex representation but are not manipulation organs, the mouth and the tongue are evolved for manipulation. Although not naturally used for pointing, the tongue constantly performs sophisticated motor control for vocalization and mastication. Such an observation points to a great potential of using the tongue and the mouth for computer input.
One difficulty in designing a tongue/mouse operated device is that the tongue has a limited range of movement. Unlike the fingers which can extend and reach, the tongue mostly stays inside of the mouth where space is limited. This makes it impractical to design a tongue device that involves a large range of movement. Fortunately, in recent years pressure sensitive isometric joysticks have been developed into very effective and small sized devices. Using the IBM TrackpointIIITM, we have designed the Tonguepoint, a tongue operated isometric input device (Figure 1).
Figure 1 Tonguepoint Device
A Tonguepoint is a mouthpiece that, similar to a dental night guard or a sports mouth guard, is form fitted to each individual's upper teeth and hard pallet. Because of this fixture the user may relax at normal jaw posture when wearing the mouthpiece. Speaking with the Tonguepoint inserted in the mouth is also feasible.
Mounted onto the mouth piece, near the roots of the front teeth, is a TrackpointTM. The shaft of the TrackpointTM is 1 cm long, and points downwards and inwards, towards the tongue tip. The shape of the handle is conical so as to facilitate a greater ability for the tongue to manipulate the handle from all directions. The tip of the joystick is composed of a soft rubber to cushion the tongue against the device. Two switches were designed for button selection. A modular bite switch can be attached anywhere between the upper and lower teeth so as to allow for button selection by biting. A second switch, linked to the outside of the mouth, is provided so as to allow for button operations by any other body organ (hand or foot). The mouthpiece is constructed of soft form fitting plastic that can be custom fitted to the individual users mouth in a about 15 minutes. This provides a secure platform for the tongue to manipulate the joystick.
Given the exploratory and speculative nature of such a design, it is important to quantitatively measure the performance of using such a device. However, since we are at an early stage of development, and there are many improvements that can be made to all aspects of the design, only a preliminary testing was conducted. Two paid volunteer subjects were recruited for pilot user testing. Neither had had previous experience with an isometric pointing device. A custom fitting Tonguepoint was made for each of the subjects. Each subject tested both a standard finger operated TrackpointIIITM and the custom made Tonguepoint device. Subject 1 tested the tongue device first and the finger device later after a short break. Subject 2 had a reversed order.
The testing task was to, as quickly as possible, select a highlighted button among an array of five by four buttons. A beep was presented if the subject clicked on the outside of the target button, and the subject was asked to continue the trial until the correct button was selected. Two tests, with 30 minutes of exploratory practice in between, were conducted with each device for each subject. Each test consisted of three sets of fifteen button selections. The button size in the three sets of testing were 15, 30 and 45 pixels respectively. The first 5 selections of each set were discarded as practice runs.
Figure 2 shows Subject 1 and Subject 2's mean selection time with both devices in each of the two tests. As can be seen, on average, the tongue device performed much poorer than the finger device in the beginning, probably due the fact that the tongue is not naturally used as an pointing organ. However, after some practice, the gap between the two devices narrowed, with the tongue performance only 5% and 57% slower than the finger performance.
Figure 2 Mean Selection Time
Due to the small number of subjects tested, modeling (Fitts' law) and statistical analysis were not suitable. However, these early results demonstrate that an isometric tongue pointing device is a promising alternative to hand input devices. More systematic experimentation is planned, pending further design improvements.
One particularly important improvement that can be made is the transfer function in the Tonguepoint. The design of the transfer function with consideration to human motor and perceptual characteristics was critical to the success of the TrackpointTM [1, 3]. The current isometric pointing device implemented in the Tonguepoint had a transfer function that was optimized for the finger, but not for the tongue. This was intentionally done so as to be able to compare the tongue and finger with as little change of other design dimensions as possible. It is hypothesized that the performance of the Tonguepoint could be further improved if the transfer function were optimized for the tongue muscles.
We thank Ted Selker for supporting the current work and building the framework under which this study was undertaken. Kim May implemented the electronic hardware used in the prototypes. These contributions are gratefully acknowledged.
1. Barrett, R. C., Selker, E. J., Rutledge, J. D., and Olyha, R. S. Negative inertia: dynamic pointing function, in Conference Companion of CHI'95: Human Factors in Computing Systems (1995), pp. 316-317.
2. Lazzaro, J. J. Computers for the disabled, in R. M. Baecker, J. Grudin, W. A. Buxton, & S. Greenberg (Eds.), Readings in Human Computer Interaction: Towards the Year 2000. (pp. 724-727). Morgan Kaufmann Publishers, San Francisco, 1995.
3. Rutledge, J. and Selker, T. Force-to-Motion Functions for pointing, in D. D. e. al (Ed.), Proc. of Human-Computer Interaction - INTERACT'90 (1990), pp. 701-705.
4. Zhai, S., Milgram, P. Buxton, W. The influence of muscle groups on performance of multiple degree-of-freedom input, in Proc of CHI'96: Human Factors in Computing Systems (1996), pp 308-315.