Ensuring Usability in Interface Design: A Workstation to Provide Usable Access to Mathematics for Visually Disabled Users

ABSTRACT

This paper presents an account of the formative evaluation of a multi-media "MATHS" workstation which is being developed to provide usable access for blind and partially sighted students reading and manipulating mathematical expressions. We argue that there is a crucial difference between notions of accessibility and usability in interface design.

Traditionally, assistive technology has been concerned with providing access to disabled users. However, unless such access embraces usability, it does not necessarily overcome the access limitations imposed by the user's disability and provide usable access. Therefore, it is essential that interface designers recognize the difference between the traditional design concept of accessibility and the more user-centered design concept of usability. The MATHS workstation is being designed with a concern for usability. In this paper, the broad context of usability is introduced. The processes of measuring usability according to ISO9241 (CD) and the development of the MATHS workstation usability requirements specification according to ISO9241 (CD) are presented. (Ed. note: These code numbers refer to a draft software usability standard prepared by the European Commission Technology Initiative for Disabled and Elderly People. They are defined more fully in the text of this article.)

We hope that this account of the application of a usability standard to the development of the MATHS workstation will be valuable to other assistive technology designers.

THE BROAD CONTEXT OF USABILITY

Accessibility is a central concept in the literature concerning technology for disabled people. One of the overriding occupational problems faced by wheelchair-users is gaining access to buildings. Many companies have attempted to ameliorate such problems by providing wheelchair ramps. Wheelchair ramps literally afford access. But, they are only a start. There are many examples of buildings with ramps (often tacked on after design, rather than included in the building specification) which are too steep for a person in a wheelchair to use with any ease at all. Other examples of "access" include buildings with relatively shallow ramps that lead to heavily resistant double swinging doors that cannot be opened by the wheelchair-user without the assistance of an attendant or passerby.

Designed to provide access? Perhaps. But hardly usable.

This distinction between accessibility and usability was acknowledged by Stig Becker of Sweden's Handikapp Institut at the First International Workshop on Access to Mathematics (IWAM -1993) when he urged designers to "make accessibility, usability". This call was echoed by John Gardner of Oregon State University, who argued that while Braille may provide access to documents for some, the need for precise, finger-tip tactile sensitivity which declines with age and the need for a considerable initial learning investment (especially for Braille mathematics notation) means that Braille can be completely unusable for many. Gardner's "DotsPlus" system of presenting tactile mathematical symbols as they appear in the printed form was designed with usability as a prime concern.

At this point it is important to make the accessibility/usability distinction clear. According to dictionary definitions, something is accessible if it can be reached. A word processing package might be accessible if the user can perform the functions contained in it. Usability encompasses much more than that. It implies consideration of the ease with which the user can perform those functions; how effectively and efficiently the user can write a paper with the package; and even the satisfaction the user derives from using it. Many writers intend to include some sense of the technology being usable when they write about it providing access. However, in line with Becker, we would suggest that usability should subsume accessibility in design discourse.

Following Whiteside et al. (1987) it can be argued that "usability" is an engineering concern. "Usability engineering" follows in the tradition of established engineering disciplines, in that its methodology is geared toward delivering useful, working artifacts. Engineering involves building things, most of which are used by people. Engineering is not carried out in avoid, but within a social and historical context. Often "the engineer" is not a person, but a multi-disciplinary team. Engineering-- delivering useful, working artifacts--is characterized by commitment to action in the world (Winograd and Flores, 1986) against a background of shared understandings. The users and the engineers must understand the requirements, including the usability requirements in much the same way. However, because of their intrinsically different backgrounds and motivations, each is capable of interpreting the requirements differently. Therefore requirement specifications must be driven by a process of negotiation involving all interested parties. Usability specifications must result in operationally defined and publicly stated criteria for satisfaction of agreed upon requirements. This is the case whether building bridges or computer systems.

So what, exactly, is usability?

In operational terms, usability is a measure of the extent to which intended users of a product achieve specified goals within a specified context of use. From the point of view of the user, usability means products that are easy-to-learn and easy-to-use. Usability means that products meet the users' needs within an overall work system, and result in greater productivity for users.

ISO9241 (CD) - A USABILITY STANDARD

The MATHS (Mathematical Access to TecHnology and Science) project is being carried out as part of the European Commission (EC) TIDE (Technology Initiative for Disabled and Elderly People) program which aims to develop a Single European rehabilitation technology market. The MATHS project is specifically concerned with addressing the mathematical access difficulties that blind and partially sighted people experience. It is intended to develop a multi-media, interactive computer workstation which enables blind and partially sighted users to read, write and manipulate algebraic expressions in a similar way to that which an exercise book allows sighted students to perform these tasks.A central focus in this project is to incorporate a concern for usability at all stages in the design and evaluation.

ISO9241 (CD) is a draft software usability standard that reflects the usability concern outlined above. It offers a framework for defining criteria for the satisfaction of specified requirements, and insists that the requirements specification process is put into context by the intended use of the system. ISO9241 (CD) insists that this requirements specification process is public and auditable. Finally it suggests that even if one construes usability as a set of characteristics of the end product, achieving usability entails a process of attending to user characteristics and needs from the very start of any software development project.

Let us consider the framework for measuring usability specified in ISO9241 (CD).

According to ISO9241 (CD), usability is a measure of the extent to which intended users of a product achieve specified goals in an effective, efficient and satisfactory manner within a specified context of use.

Effectiveness may be defined as the accuracy and completeness with which intended goals are achieved. For instance, effectiveness can be measured using the percent or number of errors, a ratio of successes to failures, or the percentage of task completed. Depending on the nature of the task, some measure of human error may need to be taken into account when measuring accuracy and completeness of task performance.

Efficiency is a measure of the amount of human, economic and temporal resources that are expended in attaining a required level of product effectiveness. Measures of efficiency include time taken to complete the task, time spent using help or documentation and subjective workload (mental effort) estimations.

Satisfaction refers to the immediate (ease-of-learning) and long-term (ease-of-use) comfort and acceptability of the overall system. Useful satisfaction measurements include the number of times the interface misleads the user or the user loses control of the system, the percent of favorable/unfavorable comments and ratings on a post-test Likert scale.

According to ISO9241 (CD) the three aspects or measures of usability (effectiveness, efficiency and satisfaction) are only meaningful within a clearly defined context of product use and are very much inter-related. Precise and accurate information about each aspect of the intended context(s) of product use must be collected before a usability requirements specification can be drawn up. The context of product use covers everything from user attributes, the equipment available for use, environmental constraints and the task goals which the product is intended to support.

User attributes include: age range, mechanical and cognitive capacities and limitations, general ability, attitude, motivation, skill, product experience, task experience, training and general knowledge. Equipment available for use relates to existing hardware and software specifications for compatibility assessment. Environmental constraints can be anything from the social environment (organizational structure, attitudes, cultureand work practices), the technical environment (or configuration) to the physical/ambient environment (workplace design and conditions). Task goals are the set of tasks that the product is intended to support, enhance or transform, and the specific goals that the product as a tool to aid the user in performing the tasks, is expected to achieve. It is important that task goals should not merely be identified. They must be clearly defined, prioritized and assigned target values.

In a usability requirements specification, a measure or a number of measures of effectiveness, efficiency and satisfaction (where relevant) for each task and in relation to each user group should be identified. This involves listing the set of tasks and types of user contexts and deciding whether effectiveness, efficiency and satisfaction should be measured by asking the user to perform that specific task (as in bench marking), by monitoring the user during free use as in logging and observation, by giving the user a questionnaire to complete, or more directly by interviewing the user.

UNDERSTANDING THE MATHS WORKSTATION CONTEXT OF USE AND USER NEEDS

The major objective of ISO9241 (CD) is to specify a usability requirements specification to guide design and evaluation throughout the duration of the development process. In order to collect the context of use and user needs data required to formulate a usability requirements specification, it is necessary to carry out some formative evaluation research.

In the MATHS project, contextual inquiry was used to analyze the overall context of use and a user survey was carried out to gather user needs. Some interview and task analysis research was also carried out to extend the contextual inquiry and user survey findings.

Contextual inquiry is a useful research tool for understanding and describing a potential user environment. It involves designers visiting potential end-user's workplace (or learning places in the case of educational products) and focusing attention on the user's needs and problems within this dynamic context. The advantage of contextual analysis is that it provides the design team with an opportunity to gain an in-depth understanding of the nature of the user's work and to consider how a product could support or even transform this work. It also helpsdevelop a sense of partnership between the designers and the users, which can be beneficial in later evaluation stages. Contextual inquiry in the MATHS project involved visiting special schools, integrated schools, mainstream schools and university disability support centers.

The results of this contextual inquiry research led to the concept of the MATHS workstation as operating in a layered context of use (Cahill & McCarthy, 1994). The MATHS workstation is intended to help blind and partially sighted students read and do mathematics. In principle it could be used in educational settings, occupational settings and for personal and home use. However, as the primary focus of the MATHS project is access to reading and doing mathematics for blind and partially sighted secondary school and third level students, from this point on,"context-of-use" refers specifically to educational settings.

The MATHS workstation context-of-use includes educational aspects, individual aspects, and interactions between the two. As each of these aspects is in itself complex, and interactions increase this complexity, there is a sense in which it can be construed as multi-layered, involving four layers: a distributed layer; a cognitive layer; a perceptual layer; and a mechanical layer.

The distributed layer concerns the social nature of the MATHS workstation in educational settings and the need for it to support student-teacher interactions and communication and cooperation between blind and partially sighted and sighted students. The cognitive layer represents the cognitive processes involved in reading and manipulating mathematics and the requirements for distribution of knowledge in the world (use of external memory) and knowledge in the head (use of internal memory). The perceptual layer concerns how visual, auditory and tactile perceptual processes differ and the resulting implications for the presentation of mathematical notation. For instance, it is necessary to insert prosodic cues in auditory presentations of mathematical expressions to compensate for the lack of spatial information available through a visual presentation. Lastly, the mechanical layer concerns the diversity of input/output needs and preferences among potential MATHS workstation end-users.

A significant finding that emerged from this research concerned the heterogeneity of the MATHS workstation potential end-user group. The different visual capabilities and limitations, input and output device preferences and mathematical abilities, etc. across blind and partially sighted users must be reflected in the design of the MATHS workstation in order to make it truly usable to a wide range of potential users.

In order to gather data about potential MATHS workstation end-users needs, a user survey was carried out (Bormans et al. 1994).The user survey or questionnaire is a list of questions that is administered to a representative group of potential users about their background, the nature of their work, the difficulties experienced in doing this work, and any feedback that they may be willing to give about their use of a product. In designing a user survey it is important to ensure that all of the questions are relevant, unambiguously stated and unobtrusive to the respondent. While multi-choice questions can be a useful way of gathering general statistical information from a large group of people, open-ended questions are often answered in a vague or meaningless manner. For this reason, contextual analysis involving some interviewing was used to obtain qualitative information about the MATHS workstation context-of-use, and a user survey was used to generate qualitative data about potential MATHS workstation end-user backgrounds, experience of mathematics and experience of computer hardware and software.

The user needs survey was administered to a large sample of blind and partially sighted students and their teachers from special schools, integrated schools, mainstream schools and third level colleges in both Ireland and Belgium. In addition to surveying blind and partially sighted students from these groups, an adapted form of the questionnaire was administered to a control group of sighted students from mainstream schools in both countries in order to determine whether specific areas of difficulty in mathematics were common to all students, or specific to those with a visual impairment.

Brief results from these surveys and the resulting implications for the design of the MATHS workstation are presented here.

First, blind and partially sighted students in Ireland and Belgium attend a range of special, integrated and mainstream schools depending on the degree of their impairments, wherethey are living in their country, and school boarding facilities.

In comparison to their sighted colleagues, only a small proportion ofblind and partially sighted students take higher level mathematics papers in formal school examinations, and in general, blind and partially sighted students find mathematics less interesting and more difficult than their other school subjects. Blind and partially sighted students report great difficulty with the mechanical (manipulation) side of mathematics, which contrasts with sighted students who have more difficulty with the conceptual (understanding) than mechanical aspect of mathematics. The implication of these general findings was that the MATHS workstation should tackle the mechanical difficulty that blind and partially sighted students experience with reading and manipulating mathematical expressions with the long-term goal of increasing visually impaired students'interest in mathematics, and in increasing the number of visually impaired students taking higher level mathematics papers in formal school examinations.

The sample of blind and partially sighted students surveyed reported finding set notation, trigonometry and logarithms as causing the most difficulty, super-and subscripts (powers and bases), tables and graphs as causing average difficulty and punctuation (brackets and commas) and algebra as causing the least difficulty. Obviously it would be ideal if the MATHS workstation could cover all of these aspects of mathematics, however in its early development stages it is only feasible for it to tackle a small subject area. While algebra is not rated as causing the most difficulty, it was felt that it presented a worthwhile area as its associated problems were mechanical rather than conceptual in nature and solutions could be provided in terms of "access" to rather than "assistance" with mathematics, which corresponds with the philosophy of the MATHS project.

In terms of specific difficulties with algebraic expressions, the students reported accuracy and speed of manipulation and memory overload as causing most difficulty with control over their navigation through an expression and confusion from linear layout of otherwise spatial representations (such as fractions) as causing some difficulty. Understanding the meaning of expressions was reported as causing least difficulty. These findings clearly support the assertion of mechanical rather than conceptual difficulties (and needs) associated with algebraic expressions, in other words "access" rather than "assistance" needs.

It was found that blind and partially sighted students tend to experience more or less the same mathematical problems as their sighted peers, except to a greater degree due to the added mechanical difficulties. This finding also supports the need for a MATHS workstation which overcomes mechanical difficulties by providing access to (rather than assistance with) reading and writing mathematical expressions.

While some of the blind and partially sighted students reported having used Apple Macintosh computers, most of the students were most familiar with IBM compatible PC's. Thus, it was recommended that the MATHS workstation be developed using an IBM PC, adapted for use by blind and partially sighted students with screen readers, speech synthesizers, Braille display devices and large character display modifications. Most of the students reported having used word-processing applications, with a small number also having experience with database and spreadsheet applications. This was interpreted as an encouraging finding from the point of view of teachers' and students' attitudes towards computer applications in the classroom and their potential interest in the MATHS workstation.

When asked about a computer-based workstation that would provide them with greater access to reading and manipulating mathematical expressions, preferences were divided between menu and command (both keyboard and voice) input. Due to the need to cater to both partially sighted and blind students and novice and expert computer users, it was recommended that the MATHS workstation provide users with access to both options. It was suggested that as with many commercial software packages, commands could be issued by choosing menu options or keyboard input of the first letter of menu options.

Most of the blind and partially sighted students said that they would be willing to invest as much learning time as required into a workstation that would overcome their mathematical access problems. This also was interpreted as an encouraging finding for the development of the MATHS workstation.

In order to extend these context-of-use and user needs findings, some interview and task analysis research was carried out as part of the MATHS workstation formative evaluation usability process.

Interviews were aimed at collecting information about the user-context, tasks to be supported and feedback on use of existing technology. An interview usually involves a one-to-one structured discussion between an interviewer (a member of the design team) and an interviewee (a potential end-user). Areas of interest can be probed in some detail and questions can be tailored to suit the interviewee.

Task analysis is a user-oriented research tool that involves collecting, analyzing and reporting information about how a task is carried out. The focus of analysis can be either in terms of the objective behaviors involved in attaining each sub-task goal, the objects and action sequences involved in performing the task, and/or the cognitive processes involved in executing the task. Information required for task analysis can be obtained from a range of sources including formal observation, informal observation, existing task documentation such as textbooks and post-task, walk-through discussion. In the MATHS project, task analysis was found to be very useful in describing how sighted, blind and partially sighted students read and manipulate algebraic expressions and how they use internal (knowledge in the head) and external (knowledge in the world) memory in performing reading and manipulation tasks.

Task analysis was also carried out as part of the development and refinement of the user-model for the MATHS workstation. This included a detailed description of the user's mathematical vocabulary, preferred cognitive strategies, goals, etc., and the kinds of functions that would be required to support them.

THE MATHS WORKSTATION USABILITY REQUIREMENTS SPECIFICATION

From the descriptions of the layered nature of the context-of-use and details of the mechanical aspects of the user's mathematical needs, it was possible to assemble a usability requirements specification according to that specified by ISO9241 (CD).

A draft version of the usability requirements specification was distributed among the inter-disciplinary project design group forr eview. This review process not only developed and refined the usability requirements specification by taking into account the ideas of a range of researchers from diverse disciplines (Psychology, Computer Science and Electronic Engineering), but it also provided a forum from which the philosophy and goals of the workstation could be finalized. Thus, it could be argued that ISO9241 (CD) is a useful inter-project group communication and decision-making tool, in addition to being a standardized process for ensuring a focus on usability in the design and evaluation stages.

As described, a usability requirements specification presents clear, detailed information about the intended users under four broad headings.First--the user attributes (general background), second--the supporting equipment (existing hardware devices and software applications and their potential compatibility), third--the user environment (the social, technical and physical aspects of the working or learning context of use) and fourth--usability measurement tasks (tasks to be supported and details of how these may be measured). Each of these, including all potential variations, were carefully and thoroughly described for the MATHS workstation intended contexts-of-use.

In terms of measurement tasks for the usability of the workstation in supporting blind and partially sighted students in reading and manipulating mathematics, a set of six representative tasks were formulated: recognition; syntactic discrimination; interpretation; glancing and browsing; manipulation; and editing. Each of these was designed to examine the individual aspects of reading and manipulation processes and to allow for specific identification of usability problems should they arise. Two further tasks were formulated to measure users' overall use of the MATHS workstation in reading and manipulating simple and complex mathematical expressions. The following list presents a summary of this set of test-tasks.

Task 1: Recognition

Definition: Identification of stimuli in the environment

Rationale: Demonstrate ability to recognize presence of algebraic notation using the workstation

Test: Read algebraic expression from workstation output/display

Task 2: Syntactic Discrimination

Definition: Ability to distinguish differences within stimuli based on their physical characteristics

Rationale: Demonstrate ability to distinguish super and sub-scripts from normal integers using the workstation

Test: Differentiate between terms such as: x^(n) +1 and x^(n+1)

Task 3: Interpretation

Definition: Accurate initial understanding of the meaning of algebraic expressions

Rationale: Demonstrate ability to classify algebraic expressions from workstation display/output

Test: Classify an algebraic expression of the form: ax^2 + bx +c = 0 (as a quadratic equation)

Task 4: Glancing and Browsing

Definition: Ability to move around a long algebraic expression, counting the number of terms and returning to specific terms

Rationale: To demonstrate ability using the workstation to glance at an algebraic expression, gain an overall impression, and also read or browse through it in a feature-discriminating manner picking out or returning to terms of interest

Test: Count the number of terms in an algebraic expression from workstation display/output and orient by reading back to a specific term

Task 5: Manipulation

Definition: Ability to move terms around in an algebraic expression in order to make it more manageable

Rationale: To demonstrate ability to move terms around within an algebraic expression

Test: Rearrange an algebraic expression so that all like terms are on one side of the equal sign

Task 6: Editing

Definition: Ability to solve an algebraic expression by adding terms, etc.

Rationale: To demonstrate ability using the workstation to move terms around in an expression and substitute new values when required

Test: Rewrite an algebraic expression by adding like terms

Task 7: Complete use of the workstation to solve a basic algebraic problem

Definition: Ability to read, interpret, manipulate and edit the solution to an unseen simple algebraic problem

Rationale: To demonstrate competence in using the full range of workstation functions to solve a simple algebraic problem

Test: Solve a simple unseen algebraic problem

Task 8: Complete use of the workstation to solve a complex algebraic problem

Definition: Ability to read, interpret, manipulate and edit the solution to an unseen complex algebraic problem

Rationale: To demonstrate competence in using the full range of workstation functions to solve a complex algebraic problem

Test: Solve a complex unseen algebraic problem

(Note: The word 'read' is used throughout the measurement task listing. It refers to reading through visual (enhanced/enlarged and standard visual display/output), tactile (Braille display/output) and auditory (speech and non-speech output) media. Steps involved in solving the tasks have not been included as the MATHS workstation does not intend to proscribe how mathematical tasks should be performed by individual users.)

It is intended that MATHS workstation evaluation will involve the collection of objective (quantitative) and subjective (qualitative) data for users' performance on each of these tasks. In accordance with ISO9241 (CD), in some cases minimum or target performance levels will be set and in other cases actual levels will be reported. In some cases, target levels will be in comparison to the use of traditional mathematical access media (such as audio cassettes) and in other cases target and actual levels will be measured in absolute terms. Thus, in addition to the set of test-tasks, it will be necessary to observe free use of the system during training and general use in order to complete the evaluation.

In order to illustrate this usability evaluation process in greater detail, the test and usability metrics for the second task: syntactic discrimination (discrimination of different syntactic types in an expression) have been presented below.

Test: If x = 2, find the value of y in the following equations:

(i) 3^(x+1) = y(ii) 4^(x) + 2 = y

Syntactic Discrimination Usability Measures:

1 Syntactic Discrimination Effectiveness
1.1 Accuracy - correct distinction of +1 as part of thes
uperscript in equation (i) and the + 2 as a constant to be
addedto the 4^(x) in equation (ii)
1.2 Completeness - correct distinctions in both tests

2 Syntactic Discrimination Efficiency
2.1 Time - time taken to perform task
2.2 Effort - workload estimation - using task load inventory

3 Syntactic Discrimination Satisfaction
3.1 Ease of use - using Likert scale

3.2 Workstation acceptability - using Likert scale

For each of these six syntactic discrimination usability metrics, for each user group (blind, partially sighted, sighted, etc.) using each component or combination of components of the workstation (Braille, speech, non-speech, etc.) and at each stage of the evaluation process (prototypes to final evaluation), measures of users' performance using the workstation to support the above syntactic discrimination task will be obtained. The levels to be reported include users' worst case, current level, best case and planned levels of performance.

Finally, it should be pointed out that the MATHS workstation usability requirements specification is intended to be a flexible technical document. As the MATHS workstation is developed, it may be necessary to add or modify the set of tasks. Additional effectiveness, efficiency and satisfaction measures for each task may also be formulated.

DISCUSSION/CONCLUSIONS

In this paper the crucial difference between the traditional and the user-centered design concepts of accessibility and usability were presented. We argued that while access involves providing a basic, but not necessarily useful, means for participation in an activity, usability ensures that this means is effective, efficient and satisfying for the potential user within the intended context of use. The emerging draft usability standard ISO9241 (CD) provides a framework and process for implementing this conception of usability at each of the three crucial stages of the design cycle. In the first stage, it involves formative evaluation and development of a usability requirements specification based on an understanding of users' needs and the intended context-of-use. In the second stage it leads to development of design recommendations and the application of the usability requirements specification to the prototype design, ongoing evaluation, and re-design. In the third stage, it involves application of the usability requirements specification to the formal final end-user evaluation. In addition to providing a usability standard and specifying a usability process, ISO9241 (CD) is a useful interdisciplinary communications tool.

ISO9241 (CD) was implemented as the usability standard for the design ofthe MATHS workstation. As ISO9241 (CD) specifies a formative evaluation of user needs and context-of-use, the first task in the MATHS project was to carry out contextual inquiry and user survey work. The results ofthese tasks were incorporated into a draft usability requirements specification which was reviewed and developed by the general interdisciplinary MATHS project group. An adopted usability requirements specification for the MATHS workstation was formulated. That will be used to design the MATHS workstation to support the users within the contexts-of-use described in this specification, and it is intended to carry out on-going and final end-user evaluation of the MATHS workstation using the tasks and metrics presented in this specification.

To conclude, the purpose of this article has been to present a discussion of the difference between accessibility and usability and to demonstrate why a concern for usability is so important ininterface design. The emerging draft software usability standard ISO9241 (CD) specifies a process that addresses usability by specifying an understanding of user needs within a defined context-of-use from the start of software development. This standard was implemented in the early stages of the on-going MATHS project in formulating the usability requirements specification, and it is expected to guide the design and evaluation of the MATHS workstation. It has also been valuable in instrumenting a sense of general commitment to the development of the one workstation within the MATHS project group. We hope that other researchers and software developers, especially those concerned with developing products for disabled users, will benefit from this broad introduction to usability and discussion of the application of ISO9241 (CD) to the design and evaluation of the MATHS workstation.

REFERENCES

Bormans, G. & Cahill, H. (1994) D1: MATHS Workstation Problem
Analysis: A Formative evaluation of the mathematical and
computer access problems as experienced by blind and partially
sighted students. EC TIDE Project 1033: MATHS Tech. Rep.

Cahill, H. & McCarthy, J. (1994) D2: MATHS Workstation Usability
Analysis. EC TIDE Project 1033: MATHS Tech. Rep.

Edwards, A.D.N. & Wesley, T.A.B. (1993) _Proceedings of the
First International Workshop on Access to Mathematics (IWAM)_,
Amsterdam. (Full reference not available.)

Whiteside, J., Bennett, J., & Holtzblatt, K. (1987) Usability
Engineering: Our Experience and Evolution. In: M. Helander (ed.)
_Handbook of Human-Computer Interaction_. North Holland Press.

Winograd, T. & Flores, F. (1986) _Understanding Computers and
Cognition: A New Foundation for Design_. Reading, Mass:
Addison-Wesley.

ACKNOWLEDGMENTS

The MATHS project was carried out as part of the Commission ofthe European Communities TIDE (Technology Initiative for Disabled and Elderly People) program. The MATHS project consortium consists of researchers from University College Cork, Ireland, University of York, UK, Katholique Universiteit Leuven, Belgium, University of Bradford, UK, Electric Brain Company, UK, F.H. Papenmeier, Germany and GRIF, S.A., France. We would like to thank all of the MATHS partners, particularly Geert Bormans and Robert Stevens, for their involvement in the user survey, their assistance in formulating the usability requirements specification and their comments on this work.