Applying the Outcomes of Needs Identification and Physical-Sensory Evaluations to Control Interface Selection

In Figure 7-9 we list specific information related to human/technology interface selection that is an outcome of the needs identification process. The information gathered reveals particular factors that should be considered during the interface selection process. For example, identifying the activity the consumer wants to perform provides us with information on how large an input domain is required and possible control interfaces to consider. If the consumer is in need of a power wheelchair and is not interested in using a computer, for example, it is not necessary to determine whether he or she can use a keyboard. Alternatively, the consumer may need to perform several functional activities (e.g., communication, mobility, and environmental control), which affects the selection of an interface. In situations such as this, it should be considered whether a different control interface for each function or a single integrated control for all the functions is to be used.

The information gathered during the physical-sensory skills evaluation gives us a profile of the user’s skills in these areas, specifically those shown in Figure 7-9. This information can be used to determine the acceptable parameters for potential control interfaces. The range measurement determines the consumer’s minimal and maximal comfortable reach and defines the geometrical requirements for the individual’s workspace. This parameter provides an indication of the possible locations for placement of a control interface (or interfaces) and the maximal distance between the extreme outer edges of the interface (e.g., the overall size of a keyboard or switch array). The resolution measurement provides data on the consumer’s ability to control his or her movement to select targets.

Given this information on the consumer’s skills, potential candidate control interfaces that have similar characteristics in terms of the number and spacing of the targets and the size of individual switches or keys can be selected. Once candidate interfaces have been selected, comparative testing is conducted. The purpose of comparative testing using the control interfaces is to provide the ATP with information on how fast the consumer can input using the control interface and the accuracy of that input. Methods for carrying out comparative testing are described in Chapter 4. During comparative testing, it is also critical that the ATP gather subjective information from the consumer on each interface that is evaluated. This information includes the ease or difficulty of use.

Control Enhancers: Interface Positioning, Arm Supports, Mouthsticks, Head Pointers, and Hand Pointers

Control enhancers are aids and strategies that enhance or extend the physical control (range and resolution) a person has available to use a control interface. In some cases a person’s control may be enhanced to the extent that he or she can select directly. In other cases control enhancers can minimize fatigue. Control enhancers include strategies, such as varying the position or the characteristics of the control interface, and devices, such as mouthsticks, head and hand pointers, and arm supports.

The person and the control interface should both be positioned to maximize function. The importance of proper positioning to maximize an individual’s function is discussed in Chapter 6. A person’s position should be observed before and during the control interface evaluation. If inadequate positioning appears to be affecting the person’s ability to control an interface, it should be addressed before continuing with the evaluation. The position of the control interface can also affect the person’s ability to activate it. Changing the height or the angle of the control interface even slightly may enhance the person’s ability to control it.

As control interfaces become more sophisticated, control-enhancing features are becoming part of the interface. For example, certain joysticks have a feature called tremor dampening that allows adjustment of the joystick for people who have tremors. Tremor-dampening joysticks are able to distinguish between tremors, which are faster and smaller, and intentional movements, which are slower and larger. The joystick is adjusted so that the tremors are disregarded and only intentional movements are detected. This adjustment enhances the ability of an individual who might otherwise be unable to operate a joystick to control a power wheelchair. A similar feature, called filter keys, is used in Windows. When the filter keys feature is activated in Windows, brief keystrokes are ignored and the rate at which keys repeat when being pressed is delayed.

Individuals who have weakness in the arm may not have enough strength to access the full range of a keyboard adequately. A mobile arm support (Figure 7-10, A), which props the arm and assists in arm movements by eliminating some of the effects of gravity, may then allow the individual to access a keyboard. For the individual who has the gross motor ability to move his or her arm and hand around a keyboard but has difficulty extending and isolating a finger to depress a key, a pointing aid may help. There are commercially available aids that can be strapped on to the hand to assist in pointing, such as the typing aid shown in Figure 7-10, B. In some cases it is necessary to custom fabricate a pointing aid for it to fit the consumer’s hand appropriately. These custom-fabricated aids can range from complex hand splints to simple tools such as a pencil with an enlarged eraser.

For individuals who lack functional movement in their arms and hands, a mouthstick or head pointer (Figure 7-11) can be used with head and neck movement to access a keyboard or perform other types of manipulation tasks (e.g., dialing a telephone number or turning pages in a book) (see Chapter 14). For a head pointer, a rod with a rubber tip is attached to a band that is worn around the top of the head. The individual can then use the end of this rod to depress keys. Besides being able to move the head vertically and horizontally, the individual must have the ability to produce a third dimension of movement to depress keys with a head pointer: forward and backward. There are also light pointers that can be worn on the head or held in the hand to control devices. One advantage of head-controlled light pointers is that it is not necessary for the user to move the head forward or backward. Light pointers are described in greater detail in the section on pointing interfaces.

Mouthsticks are often used by individuals who are quadriplegic as a result of a spinal cord injury. A mouthstick consists of a pointer attached to a mouthpiece. The user grips the mouthpiece between the teeth and moves the head to manipulate control interfaces or other objects. The shaft of the mouthstick can be made from a wooden dowel, a piece of plastic, or aluminum. In some cases, interchangeable tips for different functions (e.g., painting, writing, typing) can be inserted into the distal end of the shaft. The mouthpiece can be a standard U shape that is gripped between the teeth or a custom-made insert. Puckett et al (1988) identify a number of criteria for design of a mouthstick. Mouthsticks are also available from several suppliers. Use of a mouthstick requires good oral-motor control; later in this chapter training to develop these skills is discussed.

The consumer’s range and resolution with the control enhancer can be determined by using the same methods discussed in Chapter 4. In some cases, particularly if there is a need to extend the consumer’s range (e.g., when the head is the likely control site), it is apparent at the beginning of the evaluation that the consumer will benefit from using a control enhancer. When the user has adequate range but resolution is in question, it may not be obvious during the physical-sensory evaluation whether a control enhancer will be beneficial. In these cases it is recommended that comparative testing of candidate interfaces with and without the use of control enhancers be completed. This evaluation provides the ATP with objective data regarding the effectiveness of the control enhancer. Certain control interfaces, however, cannot be activated with a control enhancer. These include displays specially designed to be used with light pointers, eye-controlled systems, or capacitive switches requiring skin contact for activation (Lee and Thomas, 1990).

Keyboards

For written communication, a keyboard is typically considered the most efficient means of inputting information. The standard keyboard is the first choice for computer access. However, many individuals with disabilities are unable to use a standard keyboard. Fortunately, there are a number of alternatives. Table 7-2 provides examples of some commercially available alternatives to the standard keyboard.

Automatic Speech Recognition as an Alternative Keyboard

Automatic speech recognition (ASR) technology can be applied to computer access by allowing the user to speak the names of keys or key words and have these spoken utterances interpreted by the computer as if they had been typed. This approach is appealing because human speech is so rapid and voice control is so natural. ASR systems that are extremely reliable, flexible, and easy to use are available for use as full-function keyboard and mouse emulation. For example, if a word processing program is being run, then control functions such as delete, move, and print and the most common vocabulary the person normally uses, a greeting and ending for a business letter, and other similar vocabulary items can be used. If the user changes to a spreadsheet program, he or she can use vocabulary that contains items specific to that application. Microsoft Vista includes ASR as part of the built-in package of accessories.

Two basic types of ASR systems exist. With a speaker-dependent system, the user trains the system to recognize his or her voice by producing several samples of the same element. The method in which the training is handled varies among systems. The system analyzes these samples so that it can recognize variations in the user’s speech and generate a computer input (e.g., enter a given letter, a string of letters, or a control key like “return”) corresponding to what was spoken. Even after the system has been trained with several speech samples, there likely will be times when the system does not recognize the user’s speech and does not produce a response. Recognition accuracy is steadily increasing as advances are made in the computer algorithms used for analysis. Rates can be greater than 90% for general input and nearly 100% for isolated word applications (e.g., command and control, database, spreadsheet). Speaker-dependent systems can be further divided into continuous and discrete categories. Comerford, Makhoul, and Schwartz (1997) describe the development of ASR systems and describe the technical aspects of these systems.

Speaker-independent systems recognize speech patterns of different individuals without training (Gallant, 1989). These systems are developed by using samples of speech from hundreds of people and information provided by phonologists on the various pronunciations of words (Baker, 1981). The tradeoff with this type of total recognition system is that the vocabulary set is small. In assistive technology applications, speaker-independent systems are primarily used for environmental and robotic control (Chapter 14) and power mobility (Chapter 12).

Discrete speech recognition systems require the user to pause between each word for recognition to occur, which is a very unnatural type of speech. There have been reports of voice problems associated with the use of discrete speech recognition systems (Kambeyanda, Singer, and Cronk, 1997). These are due to the abrupt starting and stopping of speech required for these systems, coupled with the monotone quality required for good recognition, both of which are unnatural speech patterns. Continuous ASR systems allow the user to speak in a more normal manner, without major pauses. The rates of input are within the range of normal rates of human speech (150 to 250 words per minute). Although the possibility of damage to the vocal folds is reduced with these systems, it is not totally eliminated. Because the discrete systems are more accurate for single-word recognition, they are sometimes used for commands and control in applications such as spreadsheets and databases. Some manufacturers (e.g., Dragon Systems, Nuance, Inc., Burlington, Mass., http://www.nuance.com/) provide both continuous (e.g., Naturally Speaking) and discrete (e.g., Dragon Dictate) ASR, sometimes bundled into the same package. Currently used speech recognition systems are listed in Table 7-8. The majority of these use continuous recognition.

Speech recognition can be used for computer access, wheelchair control, and EADLs. The systems shown in Table 7-8 allow the consumer to use speech to enter text directly into a computer application program. Recognition of control words, such as “save file,” used in a word processor is also trained. System vocabulary is also growing rapidly. Early systems had recognition vocabularies (the list of words the system can recognize when spoken) in the 1000 to 5000 range. Current systems have vocabularies of 50,000 words or more. The faster speech rate, larger vocabularies, and continuous recognition all place significant demands on the speed and memory of the host computer. Continuous speech recognition systems require large amounts of memory and high-speed computers. As the cost of this added computer functionality continues to decline, these additional requirements will be less important. However, ASR systems do require more computer resources than other alternative input methods (Anson, 1999).

There are other hardware issues that are important in ASR as well. Foremost of these is the microphone. Anson (1997) discusses considerations in the choice of a microphone for ASR. Although the microphones supplied with ASR systems are satisfactory for use by nondisabled users, they are not adequate when the user has limited breath support, special positioning requirements, or low-volume speech. Most ASR systems use a standard headset microphone. Individuals who have disabilities may not be able to don and doff such microphones independently, and desk-mounted types are often used. Current ASR systems do not require separate hardware to be installed in the computer, and they use commonly available sound cards (Anson, 1999).

EADLs may also use speech recognition to access their functions (see Chapter 14). In such devices the individual can instruct the system to turn lights off and on or perform other functions by voice. The user can train the system to execute these commands with just about any sound, letter, or word.

The questions listed in Box 7-3 can be used to determine the usefulness of speech recognition for a given consumer. The key for success in using speech-activated systems is that the user be able to produce a consistent vocalization or verbalization. Differences in speech production are found not only among individual speakers but also within the same speaker. Variability in the user’s speech can cause problems with recognition. For this reason, this type of control interface may not be effective for individuals who have dysarthria. Individuals who have had a spinal cord injury and have no functional use of the upper extremities yet have good speech control are potential candidates for a speech recognition system. It is important when considering a speech recognition system to determine whether the user’s voice pitch, articulation, and loudness change or fatigue over time. Other noises or voices in the area where the speech-activated system is being used can also confuse the system, resulting in either an incorrect selection or the system having difficulty registering any selection, causing the user to repeat the vocalization several times.

Touch Screens and Touch Tablets

Touch screens are available on augmentative communication devices and laptop, palm, and notebook computers that the user activates by pointing directly to the selection set on the screen. Using a touch screen makes selection cognitively easier for many users, particularly young children, because it is more direct and intuitive. These interfaces are activated by either breaking a very thin light beam or by a capacitive array that detects the electrical charge on the finger. The electrode array used to detect where the finger, or pointer, is touching is transparent, and the touch screen can be placed over the face of a monitor. In either detection method an array of horizontal and vertical sensors is arranged so that an object the size of a finger will be detected. The position in the array determines what the interpretation of the pointing action will be, just as the specific key on a keyboard determines what the input will be. Separate touch screens can also be attached to the computer monitor or placed over a selection array on a tabletop or other flat surface. The selection set varies with the application program being used.

TongueTouch Keypad

The TongueTouch Keypad, shown in Figure 7-16, consists of nine separate small switches incorporated into a dental mouthpiece that fits in the roof of the mouth. It is a battery-operated, radio frequency–transmitting interface that activates a processor that sends IR signals to the computer. Each of the nine switches corresponds to one choice on a menu presented on a computer screen. The first menu provides choices of environmental control (e.g., television, lights), computer access (keyboard emulation), and wheelchair control. Once one of these categories is chosen, nine more choices pertaining to that category are presented, such as volume and channel control for the television; letters, keyboard array, and mouse movement directions for computer entry; or numerical choices for telephone dialing. This approach is useful for individuals who do not have motor control in their limbs but have good head, neck, and oral-motor control. In particular, the user must have good elevation of the tip of the tongue to activate the individual keys efficiently (Lau and O’Leary, 1993).

Access for Users With Cognitive Limitations

Concept keyboards replace the letters and numbers of the keyboard with pictures, symbols, or words that represent the concepts being used or taught. When the user presses on the picture, the correct character is sent to the computer to create the desired effect. As an example, a child who is having difficulty with basic arithmetic and monetary concepts may be more successful using a concept keyboard in which each key displayed is a coin of a particular denomination, rather than the value (number) or name of the coin (letters). The child can push on the coin and have that number of cents entered into the program. A simple program that asks the child to make change could be used to encourage the child to develop subtraction skills while also learning the value of specific coins. This approach is more motivating for some children and it is easier to press on a key labeled with a quarter than to enter “2” and “5.”

Very simple programs may require only two keys. For example, the SPACE key can move a cursor to different matching choices and the RETURN key can select the desired one. This concept can be used to match shapes or numbers or to control any two-choice task. It functions as a keyboard, although only two keys are used.

Another approach to concept keyboards is the use of specially designed software together with special-input keyboards. These systems do not require the use of a special input interface because they plug directly into a serial, parallel, or USB port. The software also comes with overlays for the keyboard. For example, a program to teach language concepts can be implemented by placing pictures of the concepts on specific keys and having the child generate words by pressing the correct key, causing the concept to be spoken and the picture to be repeated on the screen. When the child plays with the objects described, he or she learns to label his or her actions as well as the objects. Concept keyboards provide a direct relationship between the task and the child’s action. For example, by using a picture of the body as the “keyboard” and each body part as a “key,” a child can touch the body part when the program instructs him or her to do so. When the child does, the program can repeat the body part name and cause it to be moved on the screen. The Intellikeys (IntelliTools, Petaluma, Calif., www.intellitools.com) keyboard is often used as a concept keyboard.

An even more direct concept keyboard is the Touch Window. With this device the user merely touches the screen at the proper place and the touch screen enters the information as though it has been typed. Monitors with built-in touch screens are also available for Macintosh and Windows computers. Moving the finger on the screen can also be used to draw. This device can be placed horizontally on a table or lap tray and used as a concept keyboard with an appropriate overlay. On-screen keyboard arrays can also function as concept keyboards with the choice of on-screen elements.

There are also commercial emulation programs that reduce the complexity of the Windows environment for users who have cognitive disabilities. The Voyager suite of programs from Saltillo (Saltillo, Millersbery, Ohio, http://saltillo.buyol.com/Item/Voyager_Desktop_Suite.htm) allows individuals with cognitive disabilities to launch programs, communicate by e-mail, and browse the Web. The entire suite operates with pictures rather than words to present the user with choices removing the necessity for the user to be able to read or write. Receiving e-mail is accomplished by using text to speech to provide an auditory output. The user sends e-mail by selecting a set of pictures that enable the send function, select the recipient by picture or name, and then prompt the user to record a message and send it. Assistive technologies for persons with cognitive limitations are discussed in more detail in Chapter 10.

Eye-Controlled Systems

Often consumers use the direction of eye gaze as the only means of indicating. Manual eye-controlled communication systems have been in use for a long time. In manual systems the user communicates “yes” or “no” though eye blinks or uses the eyes to point to letters on an alphabet board to spell utterances. This manual form of using eye movement as a means of input can be automated by electronically detecting the user’s eye movements as a control interface for direct selection.

There are currently two basic types of eye-controlled systems. One type uses an IR video camera mounted adjacent to a computer display. An IR beam from the camera is shined onto the person’s eye and then reflected by the retina. The camera picks up this reflection of the individual’s eye as he or she looks at the on-screen keyboard appearing on the computer monitor. Special processing software in the computer analyzes the images coming into the camera from the eyes and determines where and for how long the person is looking on the screen. The user makes a selection by looking at it for a specified period, which can be adjusted according to the user’s needs. The EyeGaze System, Quick Glance, ERICA (Eye Response Technologies, Charlottesville, Va., www.eyeresponse.com), and Tobii (TobiiTechnology, San Francisco, Calif., www.tobii.com) are examples of two-eye–controlled systems of this type. The design principles and approach to the ERICA system are described by Lankford (2000). The other type of eye-controlled system uses a head-mounted viewer that tracks the movements of one eye. This viewer is attached to one side of the frame of a standard pair of glasses so that it is in front of one eye. The movements from the eye are viewed and converted into keyboard input by a separate control unit. One example of this type of system is VisionKey. Both types of eye-controlled systems provide the user with computer access for written or verbal communication, Internet access, environmental control, and telephone operation. To operate either type of eye-controlled system, the user must have good vision and control of at least one eye, good head control, including the ability to keep the head fairly stationary, and the cognitive ability to follow instructions.

An eye-controlled system is beneficial for individuals who have little or no movement in their limbs and may also have limited speech, for example, someone who has had a brainstem stroke, has amyotrophic lateral sclerosis, or has high-level quadriplegia. Some disadvantages of eye-controlled systems are that sunlight, bright incandescent lighting, and contact lenses may interfere with system tracking, and the cost of such systems is still rather high in comparison with other input methods. For some individuals, however, it may be the only reliable means of control.

A disadvantage of eye-controlled systems is holding the point of gaze (POG) on a target long enough for selection by a dwell time or separate switch selection. An alternative is to use an EMG control to incrementally move the cursor in small steps by muscle activations. Because the cursor only moves if a muscle is activated, holding on a target is accomplished by relaxing the muscle, and target acquisition is easier than holding the POG. However, moving large distances on the screen in small steps can be fatiguing. To take advantage of the benefits of POG for moving large distances and the EMG for narrowing down to a precise target and holding, a hybrid POG/EMG system has been developed (Barreto, Al-Masri, and Cremades, 2003). The system calculates the distance from the current cursor location to the POG of an eye tracking system. If this distance is small, then the EMG incremental stepping cursor movement is used. If the distance is large, the POG is used to move to the vicinity of the target. Trials with subjects who are not disabled showed that the hybrid POG/EMG system had faster acquisition times for targets ranging from 8.5 to 22 mm. The variance was higher for the hybrid system, indicating that additional practice and training are required to maximize its effectiveness.

Tracking of Body Features

Another approach to cursor control is the use of a camera to track body features (Betke, Gips, and Fleming, 2002). This system uses a digital camera and image recognition software to track a particular feature. The most easily tracked feature is the tip of the nose, but the eye (gross eye position, not POG), lip, chin, and thumb have also been used. The movement of the feature being tracked is converted into a signal that controls an on-screen mouse cursor. Betke, Gips, and Fleming (2002) describe the technical features of the system software in detail. Trials with nondisabled subjects in an on-screen game in which targets were “captured” by pointing the cursor at them showed that the camera mouse was accurate but slower than a typical hand-controlled mouse. With an on-screen keyboard used for a typing task, the camera mouse was half as fast as a regular mouse, but the accuracy obtained was equivalent on each system. Eleven persons with disabilities ranging in age from 2 to 58 years used the camera mouse. Eight of the 11 were able to control it reliably and continued to use it. With the increasing availability of built-in cameras in computers, the camera mouse requires only a software program to capture the body feature image and interpret its movement as mouse commands, which may make this approach more common.

Brain-Computer Interface

A significant number of people cannot effectively use any of the interfaces described in this chapter. For these individuals, the brain-computer interface (BCI) may offer promise. Although this approach is still primarily in the research stage, there are promising results to date. It is likely that we will see a much greater understanding of the biological/physical interface for the control of computers in the future (Applewhite, 2004). Figure 7-17 is an overview of a typical BCI system (Schalk et al, 2004). Features or signals that have been used include slow cortical potentials, P300 evoked potential, sensorimotor rhythms recorded from the cortex, and neuronal action potentials recorded within the cortex. The success of BCI systems depends on the type of brain signal, the methods of signal processing to extract relevant features, the algorithms that translate the features into control signals (most often a mouse-like cursor movement on the screen), user feedback, and user characteristics. BCI systems may be grouped into a set of functional components (Mason and Birch, 2003). The BCI input device provides amplification, feature extraction, feature translation, and user feedback. The control interface converts this signal to those required to control the output device (e.g., power wheelchair, EADL, computer). The device controller provides the actual control signal to the target device (e.g., signals to the motors of a power wheelchair, mouse cursor movement signals to a computer). A typical task for a user is to visualize different movements or sensations or images. An example of differing signals measured on the surface of the cortex for different imagined motor acts is shown in Figure 7-18 (Leuthardt et al, 2004). The unique signal patterns shown in Figure 7-16 can be used to generate control signals. Schalk et al (2004) give technical details of the major approaches to BCI system design. Electrodes located on the surface of the cortex have stronger and more varied signals and less interfering muscle artifact and are more stable than those attached to the scalp (Leuthardt et al, 2004). Some sensorimotor patterns that can be measured from the surface of the cortex under the skull are too weak to be measured outside the skull. The invasiveness of the electrocorticographic signals is the major drawback. In all cases, signals are mathematically analyzed to extract features useful for control (Fabiani et al, 2004).

Standard and Alternative Electronic Pointing Interfaces

The other commonly used control interface for direct selection in general-purpose computers is a mouse. There are also alternative pointing interfaces that can replace the mouse, such as a trackball, a head sensor, a continuous joystick, and the use of the arrow keys on the keypad (called MouseKeys). Box 7-4 identifies the critical questions to consider when assessing an individual for using any type of pointing interface.

It is necessary to determine whether the consumer can use the pointing interface to reach the items in the selection set (targets) and stay fixed on the target while executing the action needed to make a selection. These all imply that the selection targeted is accurate. The person may be able to get to a target area on the screen, but the size of the target may affect his or her ability to maintain that position while selecting it. Any location on the screen can be a target, and these can be of different sizes. Depending on the software program, the size of the target may be fixed or it may be possible to modify the size to meet the user’s needs. The user can use one of two techniques to make a selection. With an acceptance time selection technique, the user pauses at the selection for a predetermined period (which is adjustable) and that pause signals the selection. With the manual selection technique, the user activates another switch to let the device know that the selection has been made. The second approach provides more control for the user, but it also requires additional user motor control.

Pointing interfaces vary in terms of the tactile and proprioceptive feedback that they provide, which may affect the user’s performance. Using a pointing interface also requires a significant amount of coordination between the body site executing the movement of the cursor and the eyes following the cursor on the screen and locating the targets. The ATP should determine whether the layout of the items in the selection set is beneficial or detrimental to the user’s performance. The selection set and its layout will vary depending on the pointing interface and the software being used. It is important to know whether the layout of the selection set can be modified for a particular pointing interface and what type of modifications will benefit the user.

Modifications to Keyboards and Pointing Interfaces

There are several problems that may be experienced by individuals with a physical disability in using any of the control interfaces just described. As mentioned earlier, if a consumer is having difficulty using a particular control interface, there are three paths to pursue. A control enhancer may resolve the difficulty (e.g., when the user has limited range for accessing the interface). Modification of the interface being evaluated is another alternative, and trying a less-limiting interface is the third approach. Before a less-limiting control interface is introduced, modification of the method being evaluated should be considered. Table 7-10 lists the areas of need for which modification of a control interface may be beneficial and approaches that can be used. Each of these difficulties in using a keyboard can be addressed by either hardware or software modifications.

Selection Techniques for Scanning

The action required by the user to activate the switch to make a selection during scanning and directed scanning usually can be varied to accommodate the user’s skills. Table 7-11 lists the three scanning techniques and the level of skill required by the technique for each of the motor acts described earlier. This table is helpful in matching the scanning technique to the user’s skills. With automatic scanning, the items are presented continuously by the device at a rate that can be set and adjusted according to how fast the user can respond. When the desired selection is presented, the user selects the choice by activating the switch and stopping the scan. Automatic scanning requires a high degree of motor skill by the user to wait for the desired selection and to activate the switch in the given time frame. It also requires a high degree of sensory and cognitive vigilance for attending to and tracking the cursor on the display. With step scanning, the user activates the switch once for each item to move through the choices in the selection set. When the user comes to the desired choice, there are two possibilities for selecting it. Either an additional switch is used to give a signal to select that choice or an acceptance time is used. Step scanning allows the user to control the speed at which the items are presented. The ability to wait is not required for the scan, but it may be for the acceptance of the selection. The ability to activate the switch repeatedly, however, is highly important for step scanning. Motor fatigue can be high because of repeated switch activation.

The last technique is inverse scanning. In this type of scanning the scan is initiated by the individual’s activating and holding the switch closed. As long as the switch is held down, the items are scanned. When the desired choice appears, the individual releases the switch to make the selection. For accuracy, inverse scanning requires a high level of skill to hold the switch and release it at the proper time. Automatic scanning requires activation of the switch within a specified time frame, so inverse scanning may be easier for individuals who require lots of time to initiate and follow through with movement. Like automatic scanning, motor fatigue is reduced over step scanning because of fewer switch activations; however, sensory and cognitive fatigue is higher because of the vigilance required to attend to the display.

Many devices are capable of providing each of these scanning techniques as options for the user. Angelo (1992) performed a study with six subjects that compared these three scanning techniques. This study found that subjects with spastic cerebral palsy performed poorly with the automatic scanning technique. Step scanning was found to be the most difficult for subjects with athetoid cerebral palsy. It is helpful for the consumer to try each of these selection techniques to experience the subtle differences among them when determining which one is most suitable.

Selection Formats for Scanning

There are a number of formats in which the items in the selection set can be presented to the user for selection in scanning (Box 7-6). In a linear format, as shown in Figure 7-22, the items in the selection set are presented in a vertical or horizontal line and scanned one at a time until the desired selection is highlighted and selected by the user. With circular, or rotary, scanning (Figure 7-23), the items are presented in a circle and scanned one at a time. Because of the slowness inherent in both these types of scanning, Vanderheiden and Lloyd (1986) recommend that the array be limited to 15 choices.

To increase the rate of selection during scanning, group-item scanning can replace the single-item scan. In this case there are several items in a group and the groups are sequentially scanned. The individual first selects the group that has the desired element. Once the group has been selected, the individual items in that group are scanned until the desired item is reached. When there are a large number of items, a matrix scan can be used. In this type of scanning the group is a row of items and the items are located in columns; thus it is called row-column scanning. In row-column scanning there may be several rows of items and each complete row lights up sequentially. The row with the desired item is selected; then each column in that row lights up until the desired item is selected. Figure 7-24 shows the input required when a single switch is used with row-column scanning to produce the letter S.

There are other ways that scanning formats can be adapted to increase the user’s rate of selection. Halving is a group-item approach in which the total array is divided in halves. Each half is scanned until the user selects the desired half. The scanning then proceeds in a row-column format as described above until the desired item is reached. This same concept can be used in a quartering format in which the array is divided into fourths. Another method used to increase rate of selection is to place the selection set elements in the scanning array according to their frequency of use. For example, if letters are being used as the selection set, placement of E, T, A, O, N, and I (the most frequently used letters) in the upper left positions of the scanning array results in a significant increase in rate of selection (Vanderheiden and Lloyd, 1986). The application of these principles to augmentative communication is discussed in Chapter 11.

Coded Access

Coded access is another indirect selection input method that requires an intermediate step. As discussed earlier, one of the most common and most efficient methods of coded access is Morse code. Figure 7-25 shows the symbols for international Morse code. The required sequence of movements for obtaining the letter C is dash, dot, dash, dot. In two-switch Morse code, one switch is configured to represent a dot and the other switch a dash. Figure 7-26 shows the steps required for obtaining the letter C by using two-switch Morse code. In single-switch Morse code the system is configured so that a quick activation and release of the switch results in a dot and holding the switch closed for a longer period before releasing it results in a dash. Letter boundaries are distinguished by a slightly longer pause than between dots and dashes within one letter.

Another example of coded access is Darci code. This selection method, used with the DARCI TOO to control a computer, uses an eight-way switch code. An eight-way switch is similar to a four-position switched joystick, with the diagonal positions used as additional switch positions (Figure 7-27, A). By use of this code, the letter C is generated by moving the switch to position 2, then to position 1, and then to the center (Figure 7-27, B). It is this sequence of movements that tells the processor that the desired entry is the letter C. With this access method, it is also possible to emulate mouse movements and to access whole words. Other eight-switch (sometimes called eight-way) codes have been used in augmentative communication devices (Chapter 11).

Types of Single Switches

Numerous types of single switches are commercially available. It is also possible to custom fabricate switches, but this option is not advised for a number of reasons. Although it may seem less inexpensive to purchase the materials to make a switch, when the time it takes to make the switch is factored in, the cost involved increases significantly. In addition, custom-made switches are not as durable as commercially available switches and will not hold up over time.

When a switch is selected for an individual, it is important to consider the spatial, activation-deactivation, and sensory characteristics discussed earlier. Single switches come in many different sizes and shapes and have diverse force and sensory requirements. It is critical that the consumer has the opportunity to try out any switches being considered for a control interface. Table 7-12 summarizes the types of single-switch interfaces and gives a sampling of switches that are commercially available on the basis of the categories shown in Table 7-5.

Mechanical switches are the most commonly used type of single switch, and they can be of various shapes and sizes. Paddle switches (Figure 7-28, A) have movement in one direction. On some types of paddle switches the sensitivity can be adjusted according to the user’s needs. Wobble (Figure 7-28, B) and leaf switches (Figure 7-28, C) have a 2- to 4-inch shaft that can be activated by the user in two directions. The wobble switch makes an audible click when activated and the leaf switch does not, making the wobble switch more desirable when the switch is out of the user’s visual range, such as during head activation. Lever switches (Figure 7-28, D) are similar to wobble switches with the exception that they can only be activated in one direction. This type of switch usually has a round, padded area at the end of a shaft and produces an audible click, which also makes it desirable for activation by the head. There are also various types of button switches that come in different sizes, from a large, round switch such as the Big Red switch, to a small button switch that can be held between the thumb and the index finger, such as the Cap switch. Membrane switches consist of a very thin pad, which also requires some degree of force to activate. These pads are available in various sizes, from as small as 2 inches × 3 inches to as large as 3 inches × 5 inches. The advantages of these membrane pads are that they are flexible, can be paired with an object (by being directly attached to it), and can be used to teach the user to make a direct connection between the object and the switch. The main disadvantage of membrane switches is that they provide poor tactile feedback. This can lead to extra activations or failure to apply enough force to activate the switch. All these switches are activated by body movement that produces a force on the switch. They are considered passive switches because they do not require any outside power source. Mercury switches can be used to indicate a change in position such as lifting an arm or finger or tilting the head. This removes the need to activate a mechanical switch.

There are also switches that are activated with body movement but that do not require force or even contact with the switch. These are referred to as proximity switches. The switch is activated when it detects an object within its range. The activation range of these types of switches varies from nearly touching the switch to 3 feet away and usually the activation range is adjustable. Near switches are a series of switches that do not require contact for activation. The switches in this series use different technologies to detect the movement, from photoelectric to fiberoptic. These switches are active, meaning they require an outside power source, such as a battery, to operate.

Pneumatic switches are activated by detection of respiratory airflow or pressure and include puff-and-sip and pillow switches. Puff-and-sip switches (Figure 7-28, E) are activated by the individual’s blowing air into the switch or sucking air out of it. The individual can send varying degrees of air pressure to the switch, which provides different commands to the processor. Pillow switches (Figure 7-28, F) respond to air pressure when squeezed (such as with a hand bulb) or when pressure is applied to a cushion.

Switch Arrays, Discrete Joysticks, and Chord Keyboards

Switches are commercially available in preconfigured arrays (two to eight), and any of the single switches we have discussed can be used to design a custom array to meet the needs of the consumer. These offer the advantages of multiple signals while retaining the requirement of low resolution that is typical of single switches.

Paddle switches are often used in switch arrays when two to five input signals are desirable. A type of paddle switch that provides dual input from one control is called a rocker switch (Figure 7-29, A). A rocker switch is like a seesaw and it does exactly what it says: it rocks from side to side around a fulcrum. This design allows the user to maintain contact with the switch and perform a rotating movement with the control site to activate each side. This type of dual-switch array is often used for Morse code input, with one side signaling dots and the other side dashes. The Slot Switch (Figure 7-29, B) is one example of a commercially available paddle switch array that is already configured. The switches in this array are mounted on a base piece that has dividers between the switches. The purpose of the dividers is to help the user isolate the appropriate switch. This array is typically used with the hands or feet by someone who has gross motor skills and a fairly large range of motion. The isolation of each switch helps when the user may not be able to locate the switch visually. There are other switch arrays that are mounted and activated with the head. Switch arrays are often used for power wheelchair control; they are discussed in greater detail in Chapter 12.

At the other extreme, in terms of size, is the Penta switch array. This array consists of five switches, each approximately a fourth inch in diameter. Its overall size is 2 inches in diameter, and it is small enough so that it can be held in the palm of the hand and be activated by the thumb.

A discrete joystick is also considered an array of switches. It consists of four or five switch input signals (UP, DOWN, LEFT, RIGHT, and ENTER) that are either open or closed (off or on), with nothing in between. To close the switch, the control handle is moved in the direction of one of the other switches. Switched joysticks require limited range but moderate resolution by the user. They are available with a variety of displacements, forces, and handles to accommodate different grasping abilities of the user. If there is a maximum of five items (e.g., directions of a power wheelchair) in the selection set, the joystick functions as an interface for direct selection. When the selection set is more than five, indirect selection is required by directed scanning. Using the joystick with this method, the individual selects the direction and the device determines the speed of cursor movement.

A chord keyboard is also an array of switches or keys (typically five), each of which is intended to be pushed by one finger. Two-handed versions have 10 or more switches or keys (some have multiple keys for thumb use), and one-handed versions have five or more. The name of these keyboards is derived from the manner in which they are used for text entry. To make an entry, one or more (usually at least two) of the switches are pushed simultaneously, which is analogous to the playing of several notes together on a piano to make a musical chord. The most commonly used chord keyboard is the one used by court stenographers. With this keyboard, a stenographer can transcribe speech as it is spoken, a rate of more than 150 words per minute. For this reason, chord keyboards have often been proposed for rapid text entry by persons with disabilities. However, unless the person has good fine motor control and good coordination of the fingers, this approach is not viable. The degree of finger travel when using a chord keyboard is greatly reduced because generally only the thumb moves from key to key (usually to press a different key to change meaning of the other four keys). It would follow, therefore, that chord keyboards would reduce the incidence of repetitive strain injuries. However, the fingers still need to move to activate the keys. Like the modified keyboard layouts described earlier, there are no studies that demonstrate that chord keyboards reduce the incidence of RSIs.

The chord keyboard is used in a coded access method. Each letter, number, and special symbol is entered by pressing a combination of keys (switches). This combination is interpreted as that character by the processor. For example, to enter the letter C, keys 1 and 3 may be pressed together. The codes for each selection must be learned because it is not possible to label the keys with the necessary codes. Therefore the individual using a chord keyboard needs to have good memory skills in addition to good motor skills.

Multiple Versus Integrated Control Interfaces

A long-standing goal of rehabilitation engineers and others is the integration of systems for augmentative and alternative communication (AAC), power mobility, environmental control, and computer access (Barker, 1991; Caves et al, 1991). One of the major reasons for this emphasis is to allow the use of the same control interface for several applications, called integrated control. Integration of controls can free the individual from multiple controls and can reduce the jumble of electronic devices surrounding the person.

With recent advances in technology, it is now possible to operate several devices through one processor. The processor is capable of operating only one device at a time, and a method is set up in which the user designates the mode in which he or she would like to function. For example, there are several power wheelchairs with processors that allow the consumer to use one interface, such as a joystick, to control many functions. By selecting the drive mode, the person uses the joystick to propel the wheelchair in all directions. The person can exit the wheelchair drive mode, select the mode designated for environmental control, and turn the lights on and off in the house.

There is an inherent value in the simplification that can result from this type of integration; however, there are also many situations in which separate control interfaces (called distributed controls) and devices for each of the functions are warranted. Before deciding whether to use an integrated control or distributed controls, the implications of each method for the consumer should be carefully deliberated. As a guideline, Guerette and Sumi (1994) recommend that integrated controls be used when (1) the person has one single reliable control site, (2) the optimal control interface for each assistive device is the same, (3) speed, accuracy, ease of use, or endurance increases with the use of a single interface, and (4) the person or the family prefers integrated controls for esthetic, performance, or other subjective reasons.

In some cases the consumer may have only one body site that he or she can control, and the person may also have limited range and resolution of this control site. Trying to position more than one control interface for use by this site could be problematic. Using the same control interface for multiple functions would simplify this situation. Next, consider what is the optimal way for the consumer to operate each assistive device. Let’s say, for example, that the ATP is evaluating a consumer for control of both a power wheelchair and an AAC device. If the consumer can easily control a joystick, that would be the optimal control interface for the power wheelchair. If this is also the easiest control interface for the consumer to use for controlling an AAC device, it would stand to reason that an integrated control (the joystick) to operate both devices would be beneficial. However, if this person is able to use direct selection with an expanded keyboard for controlling an AAC device, the keyboard would be the optimal control interface for AAC. Integrating the control interfaces by using the joystick for both functions would not make sense in this situation.

Another reason to implement an integrated control interface is the user’s preference. The consumer ultimately has the final input into the selection of a control method. The consumer’s preference may be based on a sense of having better performance with one method over the other, esthetic reasons, or the importance of independence in going from one function to another. Because integrated controls combine interfaces into one unit, they typically require less hardware and tend to look better than multiple control interfaces. Integrated controls also provide increased independence for the consumer in accessing multiple assistive devices (Guerette, Caves, and Gross, 1992). The consumer does not have to depend on others to set up a different control interface or device. Some consumers place higher value on these issues than other consumers, and what is of importance to the individual consumer must be identified. There is a continuum ranging from wholly discrete systems to fully integrated systems for control interfaces, and there are advantages and disadvantages to different approaches to integration (Nisbet, 1996).

Although there are apparent advantages to using integrated controls, there may be circumstances in which distributed controls are preferred. Guerette and Nakai (1996) identify situations where integrated control may not be appropriate: “(1) when performance on one or more assistive devices is severely compromised by integrating control, (2) when an individual wishes to operate an assistive device from a position other than from a power wheelchair, (3) when physical, cognitive, or visual/perceptual limitations preclude integrating, (4) when it is the individual’s preference to use separate controls, and (5) when external factors such as cost or technical limitations preclude the use of integrated controls” (p. 64). In the case example of Mrs. Antonelli, it was easy for her to control her power wheelchair using the joystick with her left hand. However, this method was not the easiest method for her to use to operate the communication device. She had the option, however, of using another body site, and it turned out that the “best” way for her to access the communication device was by using a dual rocker switch with her right hand. If the controls had been integrated and she was to use the joystick for both power mobility and AAC, her activity output for communication would have been significantly compromised. The decision was made to use distributed controls, and her performance in communication was much improved.

In a study that measured consumer satisfaction with integrated controls, Angelo and Trefler (1998) reported that the majority of respondents indicated they were either very satisfied or satisfied with their integrated control device. An increase in independence and the ability to control other equipment such as televisions and computers were reasons the respondents gave for being satisfied with their integrated control devices. Ding et al (2003) reviewed applications of integrated controls in power mobility, augmentative communication, EADLs, and computer access. They also describe the Multiple Master Multiple Slave protocol for interfacing assistive technologies (Linnman, 1996). This protocol is an open network standard for interconnecting electronic rehabilitation devices for power mobility (Chapter 12), EADLs and robotics (Chapter 14), and augmentative communication (Chapter 11). The Multiple Master Multiple Slave standard also includes safety features that allow rapid shutdown of electronic controls (especially wheelchair and robotics) if a failure occurs. It also provides a framework for assistive technology interfaces that makes them more compatible and more easily combined into integrated controls.

Mounting the Control Interface for Use

In all situations it is necessary to address the position and placement of the control interface so that it is optimally accessed by the user. Most keyboards are connected with a cable to the computer, which allows some latitude in positioning them so they are accessible. Keyboards can be placed on stands that raise them (e.g., for mouthstick use) or easels that tilt them (e.g., for easier hand access or foot access). Some keyboards (e.g., contracted keyboards) can be mounted to wheelchairs.

It is also necessary to mount single switches, joysticks, and switch arrays in a convenient location. The most common mounting locations are attachments to a table or desk, to a wheelchair, or to the person’s body. There are commercially available mounting systems for table and wheelchair mounting. Some mounting requirements are more challenging than others. For example, it is generally more difficult to position a joystick for foot or chin use than it is to place it for hand use.

There are flexible and fixed mounting systems. Flexible mounting systems (Figure 7-30) can be adjusted and placed in various positions, which is advantageous in settings where more than one person needs a switch mounting. Costs can be controlled by using the same mounting system at different times for several people. This type of mounting system is also advantageous for individuals who require changes in the position of their control interface because of fluctuating skills or needs. The disadvantage of flexible mounting systems is that the position for the control interface must be determined each time it is put in place. Sometimes even a slight fluctuation in the position of the switch can make a significant difference in the individual’s ability to access it. Other mounting systems are fixed and are designed for use of a specific control site and switch. The advantage of this approach is that the mounting system is not as likely to move or change position and require adjustment.

Switches are also attached to the individual by straps. Attachment to the body has the major advantage of not being as affected by the person’s change in body position. If a switch is mounted to the wheelchair and the person shifts position even slightly, the switch may no longer be reachable or the new position may make it difficult to generate enough force to activate the switch.

The majority of control interfaces have a cable that connects them to the device being used. However, there are wireless keyboards, pointing interfaces, and switches. There are also separate wireless links that can be used with most switches. These links consist of a transmitter that is plugged into the switch and a receiver that plugs into the device. When the switch is pressed, the signal is transmitted to the receiver and the device. Thus the switch is not physically connected to the device. Wireless control interfaces communicate with the processor by IR signals such as those used with television remote controls. Obvious advantages of a remote control interface are that there is one less wire for the user to become tangled in and that it looks better. It can also be advantageous to have a wireless control interface when the interface is mounted on the person’s wheelchair. This arrangement allows the person to move to or away from the device being used without having to connect or disconnect the interface. In many situations the person with a disability needs a personal attendant to assist with connecting the cable of the interface to the computer. The use of a remote control interface allows the person to come and go independently, so an attendant is not needed for this task.

Speech Output

Speech is the auditory form of language, and electronic assistive technologies that provide language output rely on artificial speech. The three major applications are screen readers and print-material reading machines for persons who are blind (see Chapter 8), voice output augmentative communication devices (see Chapter 11), and alternative reading formats for persons with cognitive disabilities (see Chapter 10). The two types of speech output are digital recording and speech synthesis. They differ in the manner by which the speech is electronically produced. Table 7-14 lists the features and the typical assistive technology applications for the two approaches.

1. What is the function of the control interface? Describe the difference between a discrete and a continuous input with examples for each.

2. Define the elements of the human/technology interface and how they are related to the processor and the output.

3. What is a selection set?

4. What are the two basic selection methods used with control interfaces?

5. What are the scanning formats that can be used to accelerate scanning?

6. Why is coded access an indirect selection method? What is the selection set for Morse code?

7. What are the features included in the Macintosh universal access and Windows accessibility options?

8. What is a GIDEI, and what basic functions does it perform?

9. Explain the significance of having a USB HID specific to assistive technologies.

10. What is included in a GIDEI setup?

11. What are the relative disadvantages and advantages of software-based and hardware-based GIDEIs?

12. What does the term transparent access mean, and what features are used to implement it?

13. What is an on-screen keyboard?

14. What features are important in matching a specific on-screen keyboard to an individual’s needs and skills?

15. List three means of providing input to on-screen keyboards.

16. Examine Table 7-4. Which Morse codes listed in the nonstandard section are the same for both example systems? Why do you think these particular codes happen to be the same, given that there are no standards? Why do you think that the other codes are different for different systems?

17. What are the somatosensory characteristics of control interfaces that need to be considered in selection of an interface for a consumer?

18. Describe three control interface activation characteristics.

19. How are sensory and activation characteristics of control interfaces related?

20. What two measurements obtained from the consumer during the initial assessment provide information that will assist in identifying spatial characteristics of the control interface?

21. What is a control enhancer? List several examples.

22. Describe tremor dampening.

23. Compare the user profile for a standard, an ergonomic, an expanded, and a contracted keyboard. What user skills would lead the ATP to select one of these over the others?

24. What are the major design goals of ergonomic keyboards?

25. What are the primary considerations that would lead to the choice of speech recognition as an alternative direct selection method?

26. Describe the difference between continuous and discrete speech recognition systems.

27. What is the difference between speaker-independent and speaker-dependent automatic speech recognition systems?

28. What are the most common alternatives to a computer mouse? List at least one advantage and one disadvantage of each.

29. What factors might explain the results in the comparison study of expanded keyboard cursor control of a mouse and head pointing (Capilouto et al, 2005)?

30. What are the two most common approaches to detecting eye position and movement for use as a control interface?

31. What is point of gaze, and why is it a potential limitation in eye-tracking systems?

32. Describe the major components of a brain computer interface.

33. What are the major approaches to BCI development? Which approach do you think offers the most promise? Why?

34. List three types of modifications to keyboards and pointing devices, and give an example of the problems that each solves.

35. Describe the three different selection techniques used with scanning and directed scanning. Which one provides the user with more control and why?

36. What are the relative advantages and disadvantages of the three common scanning methods? Select a client profile that would benefit from each type.

37. Review the description of control interface flexibility in the section on characteristics of control interfaces. Pick three switches from those described in the section on selecting control interfaces, one that is very flexible, one that is moderately flexible, and one that is not flexible. Justify your choices.

38. Describe distributed and integrated control. What are the advantages and disadvantages of each?

39. What outcomes can be achieved through the implementation of training programs for development of motor skills?

40. Describe the steps taken in a training program to develop motor control.

Aaron, A, Eide, E, Pitrelli, JF. Conversational computers. Sci Am. 2005;292:64–69.

Allen, J. Linguistic-based algorithms offer practical text-to-speech systems. Speech Technol. 1981;1:12–16.

Angelo, J. Comparison of three computer scanning modes as an interface method for persons with cerebral palsy. Am J Occup Ther. 1992;46:217–222.

Angelo, J, Deterding, C, Weisman, J. Comparing three head-pointing systems using a single subject design. Assist Technol. 1991;3:43–49.

Angelo, J, Smith, RO. The critical role of occupational therapy in augmentative communication services. In: American Occupational Therapy Association: Technology review ‘89: perspectives on occupational therapy practice. Rockville, Md: American Occupational Therapy Association; 1989.

Angelo, J, Trefler, E. A survey of persons who use integrated control devices. Assist Technol. 1998;10:77–83.

Anson, D. Speech recognition technology. OT Pract. 1999;January/February:59–62.

Anson, DK. Alternative computer access: a guide to selection. Philadelphia: FA Davis, 1997.

Anson D et al: A comparison of head pointer technologies, Proc 2003 RESNA Conf: http://www.resna.org/ProfResources/Publications/Proceedings/2003/Papers/ComputerAccess/Anson_CA_Headpointers.php. Accessed June 28, 2005.

Applewhite, A. 40 years: The luminaries. IEEE Spectrum. 2004;41:37–58.

Bailey, RW. Human performance engineering, ed 2. Upper Saddle River, NJ: Prentice Hall, 1996.

Baker, JM. How to achieve recognition: A tutorial/status report on automatic speech recognition. Speech Technol. 1981;Fall:30–31. [36-43].

Barker MR: Integrating assistive technology: communication, computers, control and seating and mobility systems. Presented at Demystifying Technology Workshop, CSUS Assistive Device Center, 1991, Sacramento, Calif.

Barker MR, Cook AM: A systematic approach to evaluating physical ability for control of assistive devices, Proc 4th Ann Conf Rehabil Eng June 1981, pp. 287-289.

Barreto A, Al-Masri E, Cremades JG: Eye gaze tracking/electromyogram computer cursor control system for users with motor disabilities, Proc 2003 RESNA Conf: http://www.resna.org/ProfResources/Publications/Proceedings/2003/Papers/ComputerAccess/Barreto_CA.php. Accessed June 28, 2005.

Bear-Lehman, J. Orthopedic conditions. In Trombly CA, ed.: Occupational therapy for physical dysfunction, ed 4, Baltimore, Md: Williams & Wilkins, 1995.

Betke, M, Gips, J, Fleming, P. The camera mouse: visual tracking of body features to provide computer access for people with severe disabilities. IEEE Trans Neualr Syst Rehabil Eng. 2002;10:1–10.

Beukelman, DR, Yorkston, KM. Computer enhancement of message formulation and presentation for communication augmentation. Semin Speech Lang. 1984;5:1–10.

Blackstein-Alder, S, et al. Mouse manipulation through single switch scanning. Assist Technol. 2004;16:28–42.

Blackstone, S. The role of rate in communication. Augment Commun News. 1990;3:1–3.

Bruckert, E. A new text-to-speech product produces dynamic human-quality voice. Speech Technol. 1984;4:114–119.

Burkhart, LJ. Using computers and speech synthesis to facilitate communicative interaction with young and/or severely handicapped children. College Park, MD: Linda J. Burkhart, 1987.

Capilouto, GJ, et al. Performance investigation of a head-operated device and expanded membrane cursor keys in a target acquisition task. Technol Disabil. 2005;17:173–183.

Caves K et al: The use of integrated controls for mobility, communication and computer access, Proc 14th RESNA Conf, June 1991, pp. 166-167.

Chen, D-H, et al. Infrared-based communication augmentation system for people with multiple disabilities. Disabil Rehabil. 2004;26:1105–1109.

Chen, Y-L, et al. The new design of an infrared-controlled human-computer interface for the disabled. IEEE Trans Neural Syst Rehabil Eng. 1999;7:474–481.

Chubon, RA, Hester, MR. An enhanced standard computer keyboard system for single-finger and typing-stick typing. J Rehabil Res Dev. 1988;25:17–24.

Comerford, R, Makhoul, J, Schwartz, R. The voice of the computer is heard in the land (and it listens too). IEEE Spectrum. 1997;34:39–47.

Cook, AM. Development of motor skills for switch use by persons with severe disabilities. Dev Disabil Spec Interest Newsl. 1991;14:3.

Cook, AM, et al. Measuring target acquisition utilizing Madentec’s tracker system in individuals with cerebral palsy. Technol Disabil. 2005;17:115–163.

Cress, CJ, French, GJ. The relationship between cognitive load measurements and estimates of computer input control skills. Assist Technol. 1994;6:54–66.

Ding, D, et al. Integrated control and related technology of assistive devices. Assist Technol. 2003;15:89–97.

Drager, KD, Hustad, KC, Gable, KL. Telephone communication: synthetic and dysarthric speech intelligibility and listener preferences. Augment Altern Commun. 2004;20:103–112.

Drainoni, M, et al. Patterns of internet use by persons with spinal cord injuries and relationship to health-related quality of life. Arch Phys Med Rehabil. 2004;85:1872–1879.

Evans, DG, Drew, R, Blenkhorn, P. Controlling muse pointer position using an infrared head-operated joystick. IEEE Trans Rehabil Eng. 2000;8:107–117.

Fabiani, GE, et al. Conversion of EEG activity into cursor movement by a brain-computer interface (BCI). IEEE Trans Neural Syst Rehabil Eng. 2004;12:331–338.

Gallant, JA. Speech-recognition products. Electronic Design News. 1989;7:112–122.

Gorgens RA, Bergler PM, Gorgens DC: HandiWare: Powerful, flexible software solutions for adapted access, augmentative communication and low vision in the DOS environment, Proc 13th Annu Conf Rehabil Eng, June 1990, pp. 43-44.

Guerette, P, Caves, K, Gross, K. One switch does it all. Team Rehabil Rep. 1992;March/April:26–29.

Guerette, P, Sumi, E. Integrating control of multiple assistive devices: a retrospective review. Assist Technol. 1994;6:67–76.

Guerette, PJ, Nakai, RJ. Access to assistive technology: a comparison of integrated and distributed control. Technol Dis. 1996;5:63–73.

Horn, EM, Jones, HA. Comparison of two selection techniques used in augmentative and alternative communication. Augment Altern Commun. 1996;12:23–31.

Hortsman, H, Levine, S. Effect of word prediction features on user performance. Augment Altern Commun. 1996;12:155–168.

Hortsman, HM, Levine, SP, Jaros, LA. Keyboard emulation for access to IBM-PC–compatible computers by people with motor impairments. Assist Technol. 1989;1:63–70.

Hussey SM et al: A conceptual model for developing augmentative communication skills in individuals with severe disabilities, Proc RESNA Int 92 Conf, June 1992, pp. 287-289.

Hyman WA, Miller GE, Neigut JS: Laser diodes for head pointing and environmental control, Proc Int 92 RESNA Conf, June 1992, pp. 377-379.

Kambeyanda, D, Singer, L, Cronk, S. Potential problems associated with the use of speech recognition products. Assist Technol. 1997;9:95–101.

Kangas, K. Assessment and training of methods of access and optimal control sites. Assist Device News. 1988:5.

Kangas, K. The optimal position. Assist Device News. 1989:5.

Kasch, M, Poole, SE, Hedl, M. Acute hand injuries. In: Early MB, ed. Physical dysfunction practice skills for the occupational therapy assistant. St. Louis: Mosby, 1998.

Lankford, C. Effective eye-gaze input into Windows™. In: Eye tracking research and applications, Symposium. Palm Beach Gardens, Fl: Association of Computing Machinery; 2000:23–27.

Lau, C, O’Leary, S. Comparison of computer interface devices for persons with severe disabilities. Am J Occup Ther. 1993;47:1022–1029.

Lee, KS, Thomas, DJ. Control of computer-based technology for people with physical disabilities: an assessment manual. Toronto: University of Toronto Press, 1990.

Lesher, GW, Moulton, BJ, Higginbotham, DJ. Techniques for augmenting scanning. Augment Altern Commun. 1998;14:81–101.

Leuthardt, EC, et al. A brain-computer interface using electrocorticagraphic signals in humans. J Neural Eng. 2004;1:63–71.

Linnman, S. M3S: The local network for electric wheelchairs and rehabilitation equipment. IEEE Trans Rehabil Eng. 1996;4:188–192.

Marsden R: Personal communication, 2005.

Mason, SG, Birch, GE. A general framework for brain-computer interface design. IEEE Trans Neural Syst Rehabil Eng. 2003;11:70–85.

McCormack, DJ. The effects of keyguard use and pelvic positioning on typing speed and accuracy in a boy with cerebral palsy. Am J Occup Ther. 1990;44:312–315.

Morasso, P, et al. Towards standardization of communication and control systems for motor impaired people. Med Biol Eng Comput. 1979;17:481–488.

Murray IR et al: Emotional synthetic speech in an integrated communication prosthesis, Proc 14th Annu Conf Rehabil Eng, June 1991, pp. 311-313.

Nisbet, P. Integrating assistive technologies: current practices and future possibilities. Med Eng Physics. 1996;18:193–202.

Novak M, Olsen B: Standards work involving the general input device emulating interface (GIDEI), SerialKeys, and the universal serial bus (USB), University Wisconsin Trace Enter (2001): http://trace.wisc.edu/docs/gidei_usb/gidei-usb.html. Accessed August 15, 2005.

Phillips, B, Lin, A. Head-tracking technology for mouse control: a comparison project. In Proceedings of the 2003 RESNA Conference. Washington, DC: RESNA; 2003.

Puckett AD et al: Development of an improved mouthpiece for a mouthstick, Proc Int Conf Assoc Adv Rehabil Tech, June 1988, pp. 100-101.

Radwin, RR, Vanderheiden, GC, Lin, ML. A method for evaluating head-controlled computer input devices using Fitts’ Law. Hum Factors. 1990;32:423–438.

Ratcliff, A. Comparison of relative demands implicated in direct selection and scanning: considerations from normal children. Augment Altern Commun. 1994;10:67–74.

Salamo, GJ, Jakobs, T. Laser pointers: are they safe for use by children? Augment Altern Commun. 1996;12:47–51.

Schalk, G, et al. BCI2000: A general-purpose brain-computer interface (BCI) system. IEEE Trans Biomed Eng. 2004;51:1034–1043.

Schwejda, P, Vanderheiden, G. Adaptive-firmware card for the Apple II. Byte. 1982;7:276–314.

Shein F et al: WIVIK: a visual keyboard for Windows 3.0, Proc 14th Annu RESNA Conf, June 1991, pp. 160-162.

Shein F et al: Beyond single-switch row-column scanning with WIVKI on-screen keyboard, Proc CSUN Conference (2003): http://www.csun.edu/cod/conf/2003/proceeding/150.htm.

Simpson, RC, Koester, HH. Adaptive one-switch row-column scanning. IEEE Trans Rehabil Eng. 1999;7:464–473.

Smith, RO. Technological approaches to performance enhancement. In: Christiansen C, Baum C, eds. Occupational therapy overcoming human performance deficits. Thorofare, NJ: Slack, 1991.

Spaeth DM, Cooper RA: Designing a variable compliance joystick for control interface research, Proc RESNA 99 Conf, June 1999, pp. 131-133.

Tessler FN: The Apple adjustable keyboard, MACWORLD, November 1993.

Vanderheiden, G, Zimmermann, G, State of the science: access to information technologies. Winters JM, et al, eds. Emerging and accessible telecommunications, information and healthcare technologies. RESNA Press: Arlington, VA, 2002:152–184. http://trace.wisc.edu/docs/2002SOS-Report-IT/index.htm. [Accessed August 20, 2005.].

Vanderheiden, GC. A unified quantitative modeling approach for selection-based augmentative communication systems. Madison: University of Wisconsin, 1984. [[dissertation]].

Vanderheiden, GC, Lloyd, LL. Communication systems and their components. In: Blackstone S, Bruskin D, eds. Augmentative communication: an introduction. Rockville, Md: American Speech Language and Hearing Association, 1986.

Venkatagiri, HS. Segmental intelligibility of three text-to-speech synthesis methods in reverberant environments. Augment Altern Commun. 2004;20:150–163.

Weiss, PL. Mechanical characteristics of microswitches adapted for the physically disabled. J Biomed Eng. 1990;12:398–402.