From SAW to Comspec

1350-4533(95)00070-4 EI.SEVIER

From SAW to Comspec D. Colven*

and M. LundiXlvJ-

*The ACE Centre, Ormerod School, Waynflete Rd, Oxford OX3 8DD, UK, I-DART, BrRke &terg%d, Box 21062, Gothenburg, Sweden

Keywords:

Switch

Access

to WindowsT”

(SAW),

(:onlm~lnicatol‘

INTRODUCTION

l

The ACE Centre (Oxford, UK) and DART (Gothenburg, Sweden) have been closely involved in two major projects over the last few years. The SAW (Switch Access to WindowsTM) program is concerned with providing switch and ‘pointer only’ users with a more efficient and comprehensive access to mathematical and technical work. Comspec is examining the whole architecture of personal communications devices for people who are in need of Alternative and/or Augmentative Communication (AAC) in a more fundamental way. Comspec was prompted by the lack of flexibility and ease of use of communication aids for end users, those in need of AAC, and their supporters. At the time these two projects were commissioned, a range of devices was in use. These included: scanning typewriters; dedicated switch operated word processors; external hardware emulators for computers; content or context specific software emulators; communication software; and hardware of limited flexibility. Certainly, the last four devices listed describe the vast majority of communication devices in use even now. Most current devices are either limited electronic systems that had been created specifically for a communication task or software limited b) the speed and memory of commercial computers. This restricted environment meant that: l

Users’ requirements defined;

could not be focused or

l l

Spwificatiori

((Zonispwi

Only relatively inefficient user interfaces were available: Software access was restricted: Upgrade paths were expensive or restricted.

In other words, we had ‘digital analogue world’.

people in an

WHERE ARE WE NOW? The commercial world can now provide powerful and compact computers that are relatively inexpensive. These computers are controlled through a graphical interface with a pre-eminence of pointing (e.g. mouse) control. The operating systems offer multi-tasking environments in which programs can freely exchange data between each other. Large scale programs arc now built in a modular fashion and utilize object oriented design techniques to enhance their flexibility and reduce the complexity of monolithic code’. This more powerfIx environment is stimulating: l l

l

l

The emergence of flexible software interfaces; Greater attention to defining and implementing the users’ needs; Greater and improved access to control of and communication with the users’ environments; Closer matching of communication devices to the users’ needs.

SAW and Comspec use the power of modern commercial systems and object oriented design techniques to create flexible and efficient svstems

From

SAW to Comspec:

for people munication. THE

D. Cohen

who require

and M. Lundiilv

alternative

access or com-

SAW PROJECT’

Funding was provided by an initiative of the Training Agency (now TEED) of the Department of Trade and Industry to provide software that would aid the training and employment of people with disabilities. The project arose from the perception that switch users lose out on mainstream experiences. This lack of experience in turn hampers their educational development. There is a concentration on written or ‘spoken’ communication in the educational setting. Unfortunately, more flexible learning environments have not been available to the switch users, especially the skilled switch user who may require access to work of a mathematical or technical nature. The original concept was to create a specialized group of programs to help the switch user carry out mathematical work: writing, number handling, equation manipulation and illustration with annotated diagrams (object oriented). A review of the existing software showed that many of the functions we required were already available, but not accessible to switch users. The effort was then redirected toward providing access to these facilities. Four systems were considered for development: Macintosh, Acorn Archimedes, IBM compatible DOS, and IBM compatible MS WindowsTM. The project started at the same time as the launch of WindowsTM 3.0, and the decision to opt for WindowsTM was prompted by its superior keyboard user interface. This was also the only platform for which no switch access system was available at that time. A team of switch users in full time education was provided with machines and applications software. These end-users ranged from children to adults, all in full time education in mainstream schools, special schools and colleges of further education. Early prototypes of the SAW system were installed on the students’ machines, and frequent visits and discussions were held to advance the functionality of SAW through a heuristic approach. THE

SAW TOOL

KIT

In order to create a flexible access tool, SAW was written as a powerful design program that enables support professionals and others to create scanned systems that allow individual customization to take place whenever it was needed. In order to maximize the control efficiency, a number of features were identified. l

l

l

182

A flexible hierarchical system of scanned items (selection set). A powerful scripting language that could be programmed into any given item or group. Full emulation of all keyboard and mouse functions.

A wide range of display options for the selection sets including graphic representations, icons and symbols. The ability to ‘talk’ to co-operative applications. Close control over the desktop of running applications. A graphical design interface. A wide range of switch timing and scanning options. In the course of the development, we found that users valued flexibility and speed in text production very highly. This has made us concentrate more on text acceleration systems, prediction and abbreviation expansion”. However, flexibility to control software outside the applications we were directly supporting with a more generalized group of selection sets was also seen as important. Also, as the software and hardware associated with WindowsTM have become ever more powerful, a number of further interfaces for drawing packages, new mathematical programs, communication systems and, above all, CD-ROMs have been added. To cope with this demand for a variety of different systems, a menu is now available along with a general purpose ‘Navigator’ emulator. This desktop keyboard and mouse emulator will run any legally coded WindowsTM application. In some ways, the ‘Navigator’ (along with all generalized emulator interfaces) could be seen as a retrograde step because it decreases the efficiency of the user interface. However, users are in favour of this utility to give complete flexibility of use.

FURTHER

DEVELOPMENTS

To aid faster text acceleration, SAW 2 now includes DDE messaging that communicates with Prophet (a predictor), Abbrev (an abbreviationexpansion system) and Wordaid (an interactive alphabetic word listing system). SAW 2 can also cope properly with international versions of WindowsTM and their character sets. SAW uses a form of the TRACE Center’s ‘Access Utility’ drivers (ap-mou.drv and ap-kbd.drv), and MicrosoftTM has adopted these as standard for the special access to future versions of WindowsTM. Early tests indicate that there will be few problems in implementing SAW with the new system. We will continue to support a wide range of applications, including Write, Word and Works and Designworks in future developments. We will extend our work to cover a wide range of educational software and CD-ROMs. Multimedia systems present exciting prospects for our switch users who find searching for information in books very slow and restrictive. SAW 3 is now planned with many enhanced features, particularly in terms of graphics handling, and a greater functionality for designers and users alike. The scripting language will also be extended.

THE

COMSPEC

PROJECT

Comspec is an action research project that aims to provide a common modular software architecture as a tool to aid the production of the next generation of communication aids and assistive devices. In doing so, Comspec will try to set down some internationally acceptable standards for integrating a wide range of applications and devices which people with disabilities want to use. It will conform to software and hardware industry standards and provide both developers and users with the building blocks needed to produce seamlessly integrated solutions to powered mobility needs, environmental control systems (including ‘smart houses’), AAC devices, input and output peripherals and access to multimedia information, either locally or through the new superhighways. In short, Comspec will be the glue which will bond the mosaic of future augmentative communication aids and access systems. Comspec was inspired in 1988/89 by an IBM funded project at the Hugh Macmillan Medical Centre in Toronto”,‘. The aim of this project was to provide a set of guidelines for developers of alternative access systems. They proposed that such systems should be based on a framework of interconnecting components, all able to communicate easily with each other. This idea led the initial Comspec group to begin defining a specification for an all-purpose, modular communication device. Comspec was initiated at DART in Gothenburg, Sweden. A team was formed from the Nordic countries (Sweden, Norway, Denmark, Finland and Iceland) and the UK, and the work was funded by the Nordic Council. The first stage of the project was a review of the functionality required of a communication device. At the outset, a large number of interested profc:ssionals working in the area of communication were involved. A detailed task sheet was sent to eight national and regional centres in Europe asking them to describe the requirements of an ideal communication aid and writing tool. It was generally agreed that existing communication devices were too inflexible in the way they operated. No single device was adaptable enough to cope with the wide range of needs of people who use such devices. With children, the problem was seen as being even greater as communicators should be able to cope both with the language development of the individual and the wide range of educational situations they may experience. Most respondents were (and are) very frustrated by the absence of common standards in this small

and highly specialized field. The products of one company are usually incompatible with those of another. Devices were and are becoming more and more powerful, yet their user interfaces are different, often even when the devices originate from the same manufacturer. The wide use of non-standard components and platforms increases the cost, reduces reliability and makes bi-directional communication, for example, between a speech output device and a wheelchair controller, virtually impossible. Many saw the need for a modular and versatile design. One respondent contrasted this ideal with today’s reality of ‘the free market approach which leads to the user needing a variety of incompatible aids, thus involving much greater costs and learning time’. These results highlighted the need for development of a modular system based on software tools similar to those investigated by teams working at the Al DuPont Institute in the USA and the Hue;h Centre in Canada”. < Macmillan

THE

COMSPEC

SYSTEM

The survey was then used to guide the design of a prototype system that was written in Smalltalk, running on an Apple TM Macintosh. This prototype is public domain and copies can be obtained from the authors on receipt of two I .4 MB discs or from FTP server ftp.ifi.unit.no. The system requires IO MB of RAM to run on a Mac SE30 or better and needs 4 MB of hard disk space. Many of the lessons learned from experiences with SAW have also been applied to Comspec. It has: l l l

l l

A graphical designer interface; A modular and flexible system design; Bi-directional communication with other systems; Flexible presentation options; A wide range of representations within a given item, both m terms of text and graphics.

There are three editors that can be used to create or change a communication system. These are the Configuration, Layout and Vocabulary Editors.

CONFIGURATION In this example, we have two icon users, one operating a single switch, and the other two switches to communicate together and talk to others.

183

From SAW to Comspec: D. Cohn

and M. Lunddu

Two switch

Figure

2

The

prototype

Configuration

Editor

The Configuration Editor has as its metaphor an electronic work bench. The modules provided as ‘building blocks’ can be selected and ‘bolted’ into the system by choosing them from a menu and clicking in the work area. A ‘soldering iron’ and ‘wire cutters’ can then be used to make and break connections between these modules. It is the connections between the modules that control the flow of information through the system for both navigation and communication signals. A control panel is available for most modules that can modify their parameters, for example the tremor (post acceptance) delay for a switch or keyboard. Underlying the Configuration Editor is a kernel that controls the information passing between the components. This message management system is planned to be based on the commercial OpenDoc system’ and transmits the navigation signals that control the focus, for example, the application, a word processor and the row, column, page or paragraph, to which the communication signals are sent. It also controls the flow of the communication messages themselves. Each communication message can be represented in a number of different forms or morphologies. These are defined in the vocabulary. The kernel is intended to be an open system with well-understood behaviour that will be made available in the public domain. By these means, we hope to encourage other developers to use Comspec to construct novel systems and to expand the capabilities by creating new components. This is analogous to the way that systems such as HyperCard have developed through third

184

party commercial and public domain support. Changes to the system can be made simply and quickly, and its open nature allows new modules to be added as they become available from the Comspec or third party developers. VOCABULARY The Vocabulary Editor is the heart of the system. It contains the lexical and conceptual content of the communicator. Each element or concept in the Vocabulary Editor can be represented in a number of different morphologies so that a message can be sent to the communication partner, both human and non-human, in an appropriate form’. The forms include end user language (text), alternative language (text), other user defined languages, digitized sounds, icons and symbols including Bliss. These forms can be expanded without having to make radical changes to the system. In the future, we may wish, for example, to include video sequences as a further representation. Which representation of the message is used in any one component depends on the settings created in the configuration and the layout. LAYOUT This layout shows a screen that has been created to allow two switch users, one a single switch Bliss symbol user, the other a two switch picture communication symbol user, to talk to each other. They can also talk to others through a common text area.

W icans

Rnnrnccrntntinnc

home 03 telephone music C”P auestion

End users language

Graphical:

1 Bliss

I

Line I... Line 2.. . Figure

4

The layout is what the user themselves see. Some of the modules of the system have a visual property (for example scanned selection sets, text and icon areas, etc.), and this can be assigned to an area of the display screen and displayed in a number of morphologies depending on the users and their communication partner’s requirements. These layouts will contain layering and page properties to allow for sequencing and tree structured searching for messages. Many of the ideas have been implemented in the Comspec prototype but it is the next phase of the project that will produce the full simulator.

DEVELOPMENT

PROGRAMMEX~”

The next phase of Comspec will provide: l

l

l

l

A library of modules and development tools for the WindowsTM platform; A set of configuration tools with graphical user interfaces for use by rehabilitation professionals to create or adapt end user aids; Robust exemplars of Comspec compatible applications, for example, an environmental control system and a symbol based communicator: An introductory course and small scale evalu-

185

From SAW to Comspec:D. Coluen and M. Lund&

l

l l

ation programme involving end users and professionals; A study of the options needed for the future support and development of the Comspec system; Simple maintenance and support requirements; A network of potential Comspec developers and applications users.

Comlsffec groups :

will aim

to serve four

distinct

user

1. End users

The bottom line here is that the end user should be happy with the Comspec compatible applications they are using. Unless they wish to know more, end users will not need to know anything about the technical details of the Comspec applications they are using or how they are produced. This is because these applications (symbol editors, input options, wheelchair and environmental controllers, emulators) will be created by the System Integrators5.

Sweden and the UK. Both the Macintosh and WindowsTM 3 running on IBM compatible systems have been considered for further parallel development. Both provide powerful development environments with lightweight portable units for constructed user systems. Unfortunately, although parallel development, which will be in a form of a shortage of funds has C++, is now simpler, restricted the project to WindowsTM”. On completion of the TIDE phase of Comspec (1997), a pre-production prototype will be available for testing and evaluating beyond the project’s own quality assurance sites. By this time, it is hoped that Comspec will be extended and tested on other industrial standard software platforms such as the Macintosh. Well-supported dissemination of the software architecture to a wide and companies will also group of developers begin. We will then be able to assess whether there is general acceptability of Comspec as a practical, flexible and cost effective set of standardized tools for use in the field of rehabilitation technology.

2. Facilitators

These are teachers, non-teaching assistants, therapists and parents. They will be trained to modify the parameters, such as switch scanning speeds, of end user applications. They might also wish to learn about adapting or adding new language items to, say, a vocabulary list. This group will include some proficient end users from group 1. 3. System integrators

These people will have been trained to use the Comspec editors to create and configure new end user applications out of the available building blocks produced by system developers”. They will be able to guide support personnel and train those who wish to become system integrators themselves. Most of this group will be based in existing assessment and resource centres. Some end users may also be able to be in this group. Integrators may also provide customized tools for facilitators. 4. System and tool developers

This group will know how to use the Comspec development tools to design and implement additional modules and create specific communication and access aids. System developers may be engineers in rehabilitation technology companies, and individual developers will be at the heart of this group. Tool developers will be software engineers whose role will be to service and extend the Comspec software architecture (kernel) and set of published development tools. COMSPEC

Comspec ners in

186

-

THE FUTURE12

is now a TIDE funded project Holland, Germany, Portugal,

with partNorway,

REFERENCES 1. Wrigley I. Objects of Desire. Mac User 28 April 1995; 65. 2. Head P, Lysley A, Colven D, Poon P, Morton C. SAW (Switch Access to Windows). Anal Report on Phase 1 Oxford: ACE Centre, 1993. 3. Millar S, Nisbet P. Accelerated Wiiting for People with Disabilities. CALL Centre, University of Edinburgh. 4. McDougall J, Knysh B, Sainani D, Shein F, Lee K. Guidelines for Alternative Access System Developers. IBM Technical Report 1’X54.502 Florida: IBM Corp., Boca Raton, 1988. 5. Shein F, McDougall J, Knysh B, Sainani D, Lee K, Brownlow N, Milner M. A Model For Alternative Access Systems. Proc 12th Annual RESNA Conf; New Orleans, 1989. 6. Svanaes D, Poon P, Odor JP, Lundalv M, Comspec Working Group. Comspec Phase 1 Report. DART, Brake Ostergard, Box 21062, Goteborg (Gothenburg), Sweden, 1990. 7. Odor JP. Connecting External System to Comspec Aids-A &port to the Comspec Working Group. CALL Centre, University of Edinburgh, 1992. 8. Lund& M, Svanaes D. Comspec-Toward a Modular Software Architecture and Protocol for AAC Devices. Proc First TIDE Congress, TIDE Office, 1993; 55. 9. Lundalv M. Comspec-The Advent of an Integrated Modular Communication System. Proc Future Integrated Solutions Conf, ACE Centre, Oxford, 1994; 9. 10. Fuller P, Lysley A, Cohen D. ‘Trees in the Forest’ or ‘Seeing the Wood for the Trees’. Proc Second TIDE Congress (I. Placencia Porrero, R Puig de la Bellacasa, eds), TIDE Office, 1995; 3. 11. Svanaes D, Poon P, Odor JP, Lundalv M, Stegavik H, Torsmyr J. Comspec Phase 2 Report. DART, Gothenburg, 1992. 12. Lundglv M. Comspec-a Modular and Open Software Platform for AAC Aids on the Drawing-Board. Proc Second TIDE Congress (Placencia Porrero I, Puig de la Bellacasa R, eds), TIDE Office, 1995; 41.