The role of technology in treating clinical aphasiology has been evolving since studies first demonstrated the feasibility of using computers in the treatment of aphasic adults. This journey began with remote access to treatment in rural settings using large computer systems over the telephone. There followed the introduction and widespread use of personal computers and portable computers, with the subsequent development of complex software and multimedia programs. This changing course is not simply the result of technological progress but represents greater understanding by clinicians and researchers of the strengths and limitations of computer-aided treatment for aphasia and related disorders.

Four common types of treatment activities are appropriate for presentation on a computer: stimulation, drill and practice, simulations, and tutorials. Stimulation activities offer the participant numerous opportunities to respond quickly and usually correctly over a relatively long period of time for the purpose of maintaining and stabilizing the underlying processes or skills, rather than simply learning a new set of responses. It is easy to design computer programs that contain a large database of stimuli, and then to control variables (e.g., word length) as a function of the participant's response accuracy. Drill and practice exercises teach specific information so that the participant can function more independently. Stimuli are selected for a particular participant and goal, and therefore authoring or editing options are required to modify stimuli and target responses. A limited number of stimuli are presented and are replaced with new items when a criterion is reached. Because response accuracy is the focus of the task, the program should present an intervention or cues to help shape the participant's response toward the target response. Drill and practice programs are convergent tasks. Simulations (“microworlds”) create a structured environment in which a problem is presented and possible solutions are offered. Simulations may be simple, such as presenting a series of text describing a problem, followed by a list of possible solutions. Complex programs more closely simulate real-life situations by using pictures and sound. Simulations provide the opportunity to design divergent treatment tasks that more fully recruit real-life problem-solving strategies than those addressed by more traditional, convergent computer tasks, for example, by including several alternative but equally correct solutions. Whether computer simulations can improve communicative behavior in real-life settings remains to be tested. Tutorials offer valuable information regarding communication and quality of life to the family, friends, and others who can influence the aphasic patient's world for the better. Computer tutorials present information commonly found in patient information pamphlets but in an interactive, self-paced, format. Tutorials can incorporate features of an expert system, in which information is provided in response to a patient/family profile.

Computers can be incorporated into treatment in three fundamentally different ways. Computer-only treatment (COT) software is designed to allow patients, as part of clinician-provided treatment programs, to practice alone at the computer, without the simultaneous supervision or direct assistance from clinicians. The operation of COT programs should be familiar and intuitive for the patients, particularly those who cannot read lengthy or complex text. The program may alter in a limited way elements of the task in response to patient performance, such as reducing the number of stimuli or presenting predetermined cues in response to errors (e.g., Seron et al., 1980; Katz and Nagy, 1984). As all possible cues and therapeutic strategies that may be helpful to every patient cannot be anticipated, intervention is commonly simplistic, inflexible, or nonexistent. Consequentally, COT programs are usually convergent tasks (e.g., drills) with simple, obvious goals and, if effective, increase treatment efficiency as supplementary tasks designed to reinforce or help generalize recently learned skills.

Computer-assisted treatment (CAT) software is presented on a computer as the patient and clinician work together on the program. The role for the computer is limited to supportive functions (e.g., presenting stimuli, storing responses, summarizing performance). The clinician retains responsibility for the most therapeutically critical components, particularly designing, administering, monitoring, and modifying the intervention in response to the patient's particular needs. This relation between clinician and computer permits considerable flexibility, thus compensating for limitations inherent in the COT approach. In addition to treatment programs written specifically for use with clinicians (e.g., Loverso, Prescott, and Selinger, 1992; Van de Sandt-Koenderman, 1994), other software, such as COT word processing or a variety of video game programs, and even some web-based activities, can be used in this manner as long as clinicians provide patients with the additional information needed to perform the task.

Augmentative communication devices (ACDs) in aphasia treatment usually refer to small computers functioning as sophisticated “electronic pointing boards.” Unlike devices used by patients with severe dysarthria or other speech problems, patients with aphasia and other disorders affecting language cannot type the words they are unable to speak. ACDs designed for these individuals may incorporate digitized speech, pictures, animation, and a minimum of text. To facilitate both expression and comprehension, some devices are designed to permit both communication partners to exchange messages. Although ACDs vary in design and organization, some devices allow modification of the organization and semantic content in response to the particular needs and abilities of each patient. Researchers such as Aftonomos, Steele, and Wertz (1997) claim that for some patients with aphasia, treatment utilizing ACDs results in improved performance on standardized tests and in “natural language” (speaking, listening, etc.).

A speech-language pathologist educated in communication theory and sufficiently experienced in the clinic and in real life can create an infinite number of novel and relevant treatment activities and evaluate and modify these activities in response to unique and idiosyncratic patient behavior, even when those behaviors are unanticipated, for example, resulting from previously unacknowledged associations. In contrast, computer-provided treatment is based on a finite set of rules that are stated explicitly to evoke specific response that are (at best) likely to occur at particular points during a future treatment session, as in a game of chess. However, unlike chess, many elements of language, communication, and rehabilitation are not well delineated or universally recognized.

In describing four interrelated properties of computers and programming, Bolter (1984) helped aphasiologists better understand the relation between computers and treatment. (1) Computers deal with discrete (or digital) units of data, typically unambiguous numbers or other values, but many fundamental and recurrent aspects of communication are not clearly defined or understood. Whether during treatment or real-life, purposeful interactions, language and communication units are often incomplete, emanate (simultaneously and sequentially) from various modalities, and depend on context and past experiences. (2) Computers are conventional, that is, they apply predetermined rules to symbols that have no effect on the rules. Regardless of the value of the symbols, the sophistication of the program (“complex branching algorithms”) or the outcome of the program, the rules never change (e.g., Katz and Nagy, 1984). In aphasia treatment, all the rules of treatment are not known and those that are may not be correct for all conditions, requiring clinicians to monitor and sometimes modify rules for each patient. Unlike clinician-provided treatment, computer-provided treatment does not modify the rules of the treatment it applies; therefore, computers do not respond adequately to the dynamics of patient performance. (3) Computers are finite. Their rules and symbols are defined within the program. Except in a limited way for artificial intelligence software, unanticipated responses do not result in the creation of new rules and symbols. Therapy demands a different approach. Not all therapeutically relevant behaviors have been identified, and those that have often vary in relevance among patients and situations. Treatment software that incorporates artificial intelligence (e.g., Guyard, Masson, and Quiniou, 1990) only roughly approximates this approach, usually by reducing the scope of the task. (4) Computers are isolated from real-world experience. Problems and their solutions exist within the boundaries of the program and frequently have little to do with reality. Problems are created with the intention that they can be solved by manipulating symbols in a predetermined, finite series of steps. This lack of “world knowledge” is perhaps the most significant obstacle to comprehensive computer-provided treatment, as it limits the ability of programs to present real-world problems with multiple options and practical, flexible solutions.

In an extensive review of the literature, Robinson (1990) reported that the efficacy of computer-aided treatment for aphasia and for other cognitive disorders had not been demonstrated. The research studies reviewed suffered from inappropriate experimental designs, insufficient statistical analyses, and other deficiencies. Robinson stated that some researchers obscured the basic question by asking what works with whom under what conditions (see Darley, 1972).

There is no substitute for carefully controlled, randomized studies, the documentation of which has become the scientific foundation of aphasiology. Research reported over the last 15 years has assessed the effect of particular computerized interventions (e.g., Crerar, Ellis, and Dean, 1996) and incorporated increasingly sophisticated designs and greater numbers of subjects to assess the efficacy of computer-aided aphasia treatment, from simple A-B-A designs and comparisons of pre-and posttreatment testing (Mills, 1982) to large, randomly assigned single-subject studies (Loverso, Prescott, and Selinger, 1992) and group studies incorporating several conditions (Katz and Wertz, 1992, 1997). The efficacy of computerized aphasia treatment is being addressed one study at a time.

See also speech and language disorders in children: computer-based approaches.