T.V. Raman is the author of Emacspeak and Auditory User Interfaces: Toward the Speaking Computer. He is currently a computer scientist at Adobe Systems. He can be reached at [email protected]
The primary function of a computing application is to accept user input, perform any necessary computations, and display the results. Thus, the role of the user interface (UI) is to accept user input and display results.
User interface is a means to an end -- its role is to facilitate man-machine communication. A truly well-designed UI is therefore one that the user does not notice. However, today's computer interfaces have become so complex that they often overshadow the computing task that they were supposed to enable.
Speech Access To Computing
Alan Perlis once said, "One man's constant is another man's variable." Today's computer interfaces certainly reflect the truth of this statement, especially when one tries to provide alternative interfaces such as speech access to standard computing tasks.
Motivated by the needs of adaptive technology, attempts at providing speech access have until now concentrated on retrofitting speech interfaces onto a visual UI that was designed with no thought to alternative modalities. These interfaces are cumbersome, and have been tolerated primarily because users possess no alternatives. The inadequacy of these screen-access interfaces points to the need for separating the UI from the underlying computation engine and treating speech as a first-class medium in the interface.
In a world solely dominated by Graphical User Interfaces (GUIs), application logic has become irrevocably tangled up with the user interface. Thus, even applications developed in modern object-oriented languages like Java suffer from a distinct lack of separation of concerns -- this is evinced by the raging debates over the various Java user interface toolkits such as AWT, IFC and AFC. If user interface and application logic were clearly separated, the choice of UI toolkit would not in the least affect the rest of the software application.
Ubiquitous software needs to do more than just run anywhere; it needs to be usable everywhere. The need to expose the user-level functionality provided by these applications via a multiplicity of appliances such as mobile phones and smart cards, as well as traditional computers, offers software engineers a unique opportunity to rethink application design.
Separation Of Concerns
The speech-enabling approach described in my book, Auditory User Interfaces: Toward the Speaking Computer, benefits greatly from the separation of the computational and user interface components of software applications -- a concept familiar to researchers in the field of programming methodology as Dijkstra's classical separation of concerns. Application designers can implement desired functionality in the computational component and have different user interfaces expose the resulting power in the manner most suited to a given user environment.
This separation of computation and user interaction is significant in both software architecture and the evolution of tomorrow's personal communication devices. The size and shape of today's smallest laptop is determined by the user interface peripherals -- the keyboard and visual display. But a computing device in itself needs neither of these to perform its computations. As computers become more prevalent in our day-to-day environment, the current design of computational devices would force us to have several visual displays, numerous keyboards, and a plethora of mice in our offices and living rooms.
Separating the computational component from the user interface enables smart personal communication devices to share a common set of peripherals. Thus, instead of every computing device coming with its own display, intelligent communication devices of the future could share a single high-quality flat-panel visual display that hangs on our walls. Such an evolution would vastly speed up the convergence of computing and telecommunications. Similarly, a computing device that can recognize and synthesize speech could be designed to fit in a pocket! Such dedicated communication devices could then be the ears and voice of all personal appliances.
The benefits of separating the UI from the underlying computation engine will be first felt in the field of adaptive technology. But in the long-term, the benefits of adaptive technology are far greater to the general user. History has proven this over and over again -- how many of us remember that the invention of the telephone was a byproduct of attempts to invent a hearing aid?
Related web sites
- Emacspeak: A Speech-Enabling Interface (Dr. Dobb's Journal, September, 1997)
- Auditory User Interfaces: Towards the Speaking Computer (Kluwer Academic Publishers, August 1997, ISBN 0-7923-9984-6)
- Emacspeak home page
These op/eds do not necessarily reflect the opinions of the author's employer or of Dr. Dobb's Journal. If you have comments, questions, or would like to contribute your own opinions, please contact us at [email protected].