Future intelligent telephone terminals: method for user interface evaluation early in the design process

Future intelligent telephone terminals

Dikmans, L. (1994). Future intelligent telephone terminals: method for user interface evaluation early in the design process. (IPO-Rapport; Vol. 982). Instituut voor Perceptie Onderzoek (IPO).

Document status and date: Published: 19/05/1994 Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:

openaccess@tue.nl

,'-:

,

Institute for Perception Research 19.05.1994

P.O. Box 513 - 5600 MB Eindhoven

Rapport no. 982

Future intelligent telephone terminals Method for user interface evaluation early in the design process

Future intelligent telephone terminals

Method for user interface evaluation early in the design

process

Graduation Thesis

Lonneke Dikmans

1994

Department of Cognitive Science

Katholieke Universiteit Nijmegen

Abstract

The telephone is an important device in business organizations. The literature shows that users experience difficulties operating conventional telephone terminals. A prototype of a screen based telephone interface was developed using PAID (Philips Advanced Interactive Display). With the prototype users can make a call, make and send notes, and make drawings together with the person they are calling. Users can also keep a list of names and telephone numbers in the telephone with which they can make a call. The prototype was evaluated to determine whether such an interface is fit to support the activities of secretaries and can solve current usability problems. The evaluation was done early in the design process and the feasibility of such an early evaluation is discussed. It was shown that users prefer to make calls with the names list and that they prefer to work in the current mode when possible. It was demonstrated that it is feasible to do user test very early in the design process. The set-up and the results found can be used in successive phases of the design process.

Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Multi-tasking and interruptions in the office environment . . . . . . . . . . . . . . . . . 2

1.2 The terminal device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.3 Device technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.5 The experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.1 Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.2 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.3 Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.4 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.5 Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.6 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2. 7 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.1 General Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.2 Call . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

3.3 Adding names to the directory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.4 Edit Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.5 Mail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.6 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.1 Early testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.2 The experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4.3 The analysis method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Appendices Background form Questionnaire Instruction Taskdescription Activity nets A B C D E

Chapter 1

Introduction

The telephone is an important communication device in business organizations. A study (Brouwer-Janse, Scheffer, Vissers & Westrik, 1992) was conducted to investigate the pattern of activities of secretaries in the working situation and the role of the telephone as a communication device. It was found that the activities of secretaries are determined by certain constraints:

• The activities of their bosses and their department determine the workflow of secretaries. • They cannot control the execution or completion of their tasks because their work

environment is open to interruptions.

• Most of the time they are engaged in multi-tasking, they very seldom do one task at a time.

The various activities of secretaries can be represented in a structure of three layers with embedded activities in each layer. The first layer consists of long lasting activities, for example word processing. The second layer consists of activities that deal with

communication (like a phone call), and the third layer consists of activities that support the activities in the first and second layer (for example calendar management or data search and retrieval).

Telephones provide, apart from simple functions like making a call, several functions to increase the efficiency and flexibility of secretarial tasks. These functions include for example, placing calls for a boss, short-code dialing, last number redialing, auto ring back, follow me, conference call and hands-free dial mode. Apart from short code dialing these functions were hardly used and some were even unknown to the secretaries (for example, follow-me) (Brouwer-Janse et al., 1992).

There are several technologies available that could improve the effectiveness of

telephones to support the activities of users. These technologies include, among others, speech recognition and speech synthesis, calendar management, pen based computers, character recognition, and in the future the services enabled by the integrated services digital network (ISDN). However, little is known about the optimal way to implement novel user interfaces with these new techniques.

The objectives of the overall project of which this study is a part are:

• Create an experimental environment in which novel technologies for telephone communication can be generated and evaluated with people using systems/terminals during an extended time period.

• Have the facilities to test user interface concepts in a very early stage with the possibility of varying tasks and users.

• Test prototype systems in-house before they are evaluated in the field with customers. A prototype of a new telephone interface was developed by the Systems Project Centre of Philips Research Laboratories using PAID (Philips Advanced Interactive Display). The purpose of the study reported here is:

• Review relevant literature about multi-tasking, usability of telephone terminals, device technology like pen computers, and evaluation methods.

• Development of realistic experimental tasks that take into account the working situation of secretaries.

• Development of an experimental procedure that is fit to be used early in the design process.

• Investigate whether this interface is useful for future intelligent telephone terminals that make use of integrated voice and data transmission.

In order to be useful, the telephone terminal has to support the activities of secretaries, should be easy to operate for people who are used to standard telephone sets, and should be more efficient than standard telephone sets.

In the literature several issues concerning the usability of the terminal device, the

advantages and disadvantages of pen computers and touch screens, and multi-tasking are addressed. These issues are reviewed in the following sections, and the experiment that was conducted to test the usability and initial acceptance of a screen based telephone interface is discussed in chapter two, three and four.

1.1 Multi-tasking and interruptions in the office environment

The study that investigated the pattern of activities of secretaries (Brouwer-Janse et al, 1992) showed that they are often engaged in more than one activity at one time and that their work environment is open to interruptions. Therefore the telephone interface has to support multi-tasking (the execution of multiple tasks at the same time by the user) and minimize the effect of interruptions. Some guidelines and experiments are reviewed in this paragraph.

Studies (Callier & Eyrolle, 1992) that have been devoted to performance in time sharing task situations have shown that:

• Perfect time sharing, i.e. no degradation of performance of either task, only occurs when the tasks can be processed automatically by the person. (Shiffrin and Schneider, 1977) • There is generally some interference between the two tasks.

• Serial rather than parallel processing is generally utilized in time sharing task situations. According to Cypher (1986) problems can arise when activities involve working with a computer, because there are often mismatches between computer programs and user activities:

• If a single activity of the user calls upon more than one application.

• If more than one activity calls upon one program. The problem is that every activity has its own context, but the program only has one set of context variables so a context clash occurs. A solution could be to call upon a separate instantiation of the program for every activity, but this would cost too much time.

• In case of 'while-am-at it' activities. This occurs when users are performing a certain task (for example, reading e-mail message number three) and they interrupt this task to do another task in the same application (for example, forward message number six). The two activities share exactly the same context but their goals have nothing in common. This can be confusing because the system state can change as a result of the second task (for example, when using the 'next' command, message number seven is selected instead of message number four), without the user being aware of it. A solution for this problem could be to make sure that small operations do not change the entire context.

Effect of interruptions

Callier and Eyrolle (1992) conducted an experiment to investigate human performance when processing of one task is interrupted by the need of a second task. They tested the effect of three independent variables on inter response interval and error rate. The

independent variables were time constraint, complexity, and similarity. The task consisted of selecting certain items from a scrolling display with alphanumeric items. For example, subjects had to select all even numbers and vowels. Time constraints were varied by varying the scrolling rate. The complexity was defined as the amount of information that had to be processed by the subject and similarity was the amount of overlap between the nature of the items and the rules to be applied.

It was found that interruption of one task in order to carry out another task led to a longer processing time and higher error rates. Two types of errors could be distinguished: errors that occurred both in single task situations and in switching task situations and errors that mainly occurred in the switching task situations. The errors of the second category

consisted of intrusions, confusions and omissions.

Gillie and Broadbent (1989) also examined the effect of interruptions. They investigated the effect on the completion time of four variables:

• The length of the interruption.

• The similarity between the main task and the interruption task. • The complexity of the interruption task.

• The need for immediate attention to the interruption task versus the opportunity to rehearse the state of the main task.

The main task consisted of a computer-based adventure game. The interruption task in experiment 1 and experiment 2 was simple mental arithmetic, such as adding two digits. In experiment 3 the interruption task was free recall, in experiment 4 the mental arithmetic was complicated by coding the digits as letters.

It was found that the nature of the interruption, in terms of similarity to the main task, and the complexity of the interruption, in terms of the amount of processing or memory storage required, seem to determine which interruptions will be disruptive and which will not. The length of the interruption and the opportunity to control the point at which the main task was stopped were important factors in determining how disruptive the interruption was.

Guidelines

In the literature several guidelines to support multi-tasking and to minimize the effect of interruptions are mentioned. Systems that are designed to support multi-tasking should meet the following conditions (Myata & Norman, 1986):

• The system has to be designed in such a way that it is easy to interrupt an activity at any desired time.

• Sufficient information has to be saved so that when the suspended activity is resumed it can be continued where it is left off.

• There has to be a reminding structure to remind users of the fact that a certain task is not completed.

development of an uniform interface system which should handle the coordination and execution of activities at many levels:

• Reducing the mental load when switching tasks. Today, users keep a notebook and next to their computer to make notes when they come up with a novel idea. The interface should allow the user to make notes everywhere and when the user is ready to spend time on the other task the note can be named, and relocated. Unfortunately, they did not test this, so it is possible that users get confused by this possibility.

• Suspending and resuming activities. Users rarely complete any time-consuming activity before beginning with another task. So there has be to a mechanism for easy

suspending of tasks.

• Maintaining records of activities. When the system keeps a command sequence activity script it can be reused when a comparable situation occurs.

• Functional grouping of activities.

• Allow users to have multiple perspectives on the work environment. (for example, temporal ordering or goal ordering)

• Support the ability to have multiple instances of a particular file or note.

Reminders

People can use reminders to make sure that they can concentrate on the execution of their task instead of on their planning. Reminders are necessary when suspended activities have to be resumed at the right time and the right place. According to Miyata and Norman (1986) there are two aspects to reminders, reminders as a signal and reminders as a description. A reminder as a signal indicates that something is to be remembered. A reminder as description aids in retrieving what is to be remembered. A cue is most effective if there exists discrepancy between the task and the cue (for example, a visual task and an auditory cue).

Reminders should inform the user when conditions are ready for resumption, remind the user when something has to be done immediately, not distract the user from the current activity, and they should list activities that are currently under suspension.

The timing of the reminder is important, because reminders are interruptions as well (Miyata & Norman, 1986). User activities can be divided into seven stages : Establishing the goal, forming the intention, specifying the action sequence, executing the action

sequence, perceiving the system state, interpreting the system state and evaluate the state with respect to the goals and intentions (Norman, 1986). Reminders interrupt most during planning, execution, perception and evaluation and least after evaluation and between execution and evaluation (Miyata & Norman, 1986).

1.2 The tenninal device

Users experience difficulties operating conventional telephone terminals and do not use the wide functionality telephone terminals offer. Reasons for these usability problems and guidelines to prevent them will be discussed in this section. After that the use of function keys, and mnemonic aids as a solution are reviewed and issues that have to be

considered when designing multiservice terminals are addressed.

There are many other possibilities in addition to making and receiving calls for telephone services, for example displaying the number of incoming calls or distinctive ringing for certain people's phone numbers (Mitchel & Todd, 1985). With the arrival of ISDN, other services will become available to telephone users, such as the single subscriber call number (Noe, 1988). This means that users no longer have a special fax number, but that

they can sent and receive facsimile on their 'normal' telephone number. However, people will only use existing and new special features if they are easy and effective to use.

Usability problems

Experiments have shown that users experience difficulties when operating services such as call forwarding with the twelve push-buttons on standard telephones. All the services have to be operated with the ten digits and the '#' and the '*' button. Several reasons are mentioned in the literature to explain these difficulties. For example the conceptual model users have of the device (Bennett & Klinger, 1990) or the way the features are presented to the users (Neumeier, 1990). These and other reasons are discussed in the next sections.

General principles

According to Neumeier (1990) most subscribers of modern services don't use the wide range of functionality of their telephones because the functions are not adequately presented to the user. According to him the interface should follow certain general principles to avoid usability problems:

• Restriction to a small amount of useful features, because the interface gets more complicated when there are more features.

• Guiding the user by clear visual feedback.

• Reduction of the number of keys. When a 'one feature - one button' concept is used, it becomes difficult to find the right button at the right time.

Conceptual mode/1

When people use a system, for example, a car or a computer, they have a conceptual model of that system. They use this representation when they interact with the system to predict the results of their actions, to evaluate these results and to solve problems.

According to Kellog and Breen (1987) the ability of a person to use a system is related to their conception of the system. The closer a user's conceptual model is to the system model, the better the performance becomes.

Bennett & Klinger (1990) interviewed professional telephone users and found that they had a conceptual model of telephony that was inconsistent with the interface design. They suggest that this is the reason that supplement features are hardly used. Two prototypes were developed that were based on a personal face-to-face visit analogy in order to establish a better match between the interface and the user's conceptual models of

telephony. The default call type in these prototypes was conference call. When a person is speaking to someone and a second call arrives, answering the call will establish a three-way conference call. Anyone can leave a conference at any time, but nobody can force anyone else to leave. They did some usability tests with Bell Labs employees and most subjects found the prototype easier to use than the traditional interface. It was not tested whether there was a change in conceptual models of telephony as a result of using the prototype.

To make sure that users will be able to operate the extra services that ISDN offers, the interface has to be consistent with users conceptions of telephony.

1

There is a distinction between the term 'mental model' and the term 'conceptual model' (Sein & Bostrom, 1989): a mental model is an internal representation of the system that provides predictive and explanatory power to the user, a conceptual model is a basic depiction of the system, external to the user. The term conceptual model is used here, because the authors mentioned in this section use this term.

Function keys

When designing a telecommunication terminal one has to decide which functions and function keys are going to be implemented. Function keys can be divided in three groups: fixed keys (like <return>), programmable keys (the function of which can be filled in by the user) and soft keys (the function of which is determined by the system application).

The following guidelines are important when function keys are going to be used (Frankhuizen, 1985):

• Consistency of the function key. Pressing a key in one situation should have the same effect as pressing that key in another situation.

• The number of keys. More and more telecommunication possibilities are offered to users. When there is a different key for every application the interface has to be changed entirely every time a new service is added.

• Information presentation. Avoid computer terminology, use known sets of symbols on the keys, and always give information about the result of pressing a function key.

• Fixed keys versus soft keys. Keys should only be fixed when their meaning is applicable to a wide variety of services.

Mnemonic aids

When users want to establish a certain goal (for example, transferring a call) they have to make two connections: 1) between the goal of their action and the specific command needed and 2) between this specific command and the key presses that instantiate it. Mnemonic aids can make it easier to establish these links. Egly, Jeffries, Leban, Loebner, Parker & Sears (1985) distinguish four classes of mnemonic aids: metaphoric visual aid, iconic visual aid, character aid and grammatical aid. They designed instances of these mnemonic aids for a voice message system. The metaphoric visual aid provided a spatial representation of the system functions. They used the features of a buccaneer's face, for example send message is the far ear + open eye. The iconic visual aid provided a

functional representation of the system functions. They used objects on a desk, for example send message is the out tray + the filling cabinet. In the character mnemonic commands are constructed according to the first letter of each word of the command. For example, send message is 'm' because the command is mail message and the word message is to be ignored. In the grammatical mnemonic commands are constructed from words associated with particular telephone keys. For example, send message is 'Go'. They did not test the different classes of mnemonics.

Root & Koster (1986) developed an interface which was extensible, easy to use and easy to memorize by using a mnemonic syntax for several services, for example, speed dialing. Speed dialing enables the user to store telephone numbers which can later be dialed with one or two preselected keystrokes. They conducted an experiment that compared an interface with numeric syntax to an interface with (character-) mnemonic syntax. For example, in the interface with numeric syntax the number to forward calls was 103 {phonenumber} and to undo the forwarding, the number was 104. In the interface with mnemonic syntax the code to turn call forwarding on is CF# {phonenumber} and the code to turn it off is *C CF#. 'CF#' stands for call forward and '*C' stands for cancel. The subjects in the experiment were members of the Network Services Research Division of Bell Communications Research. Users of the mnemonic interface remembered more services and service commands than users of the numeric interface, but they did not change their pattern of use as a result of the new interface.

A keypad interface does not provide the opportunity to display entries in the speed dialing list, so users have to resort to memory to manage the contents of this list. At Bell

Communications Research users have a variant of speed dialing: memory dialing. Memory dialing allows users to associate a telephone number with a string of arbitrary length. Root

& Chow (1987) used a terminal based telephone interface with the following editing

services for memory dialing: display entries, add/delete entries, change a dialing code and change a telephone number. These functions can be accessed by using the terminal keyboard. An enhanced terminal provided the same functions plus a "call-number" command for placing memory dial calls directly from the terminal instead of from the telephone keypad. It was found that the use of memory dialing increased during the

experiment. Users employed both the terminal and the telephone interface. Each modality, however, was used for different aspects of the task: the terminal was used for information management; i.e. creating and modifying entries, and the telephone was used for making calls.

Multiservice tenninals

Multiservice terminals are terminals that combine different communication services, such as speech and text communications. According to Noe (1988) these terminals offer users benefits, such as:

• Concurrent processing of, for example, telephoning and data retrieval. • Simultaneous voice and data communication.

• Economy of space, because only one device for different services is necessary.

• Less learning effort than for single service terminals, because only one device has to be learned.

To avoid usability problems, however, the interface should be an integrated, standardized system. For example, function keys and displays have to be designed according to the same principles for all services. Furthermore, speech mechanisms are needed to prevent the interference of messages to users with other user tasks. According to the German DIN standards the ergonomic requirements for the dialog design of the user interface for multiservice terminals are: suitable for the task at hand, self describing, controllable, conform to expectations, and robust (Noe, 1988).

Two types of control dialogue for multiservice telephones can be distinguished (Damay & Poulain, 1985): the dynamic dialogue and the permanent dialogue. The dynamic dialogue is based on principles used in ergonomic psychology. The design of the terminal is independent of the facilities that are initially available. Characteristics of the dynamic dialogue are:

• When not in use, the keyboard shows none of the facilities available.

• The user chooses a range of functions by keying one of the permanent keys (for example, call).

• A list commands that are applicable is displayed on five soft keys. • The same label always appears in front of the same key.

• After each command, the user receives feed-back or a query from the interface.

The permanent dialogue is based on paradigms from experimental psychology. The design of the terminal is based on a classification of functions of the telephone services. The lay-out of the keyboard is derived from this classification. The keyboard offers a permanent

visualization of the main categories (for example, operations on communications or tax options) of actions. Damay and Poulain tested the dynamic dialogue and the permanent dialogue terminals with average telephone users (people that make less than five calls a day). The following variables were measured: success/failure, execution time, thinking time (the time before the first key is typed), and the number of key presses. The dynamic dialogue was superior to the permanent dialogue with respect to the average number of successes. It was also found that with the dynamic dialogue there was progress for all the subjects at each session with respect to all the measures and for all the scenarios. With the permanent dialogue there was a wide dispersion of performances between subjects and between scenarios and there was a decrease of performance after the instructions of use were taken away. The following design rules for dynamic dialogues were derived from this experiment:

• Take the learning phase into account and make the dialogue explicit and simple. • The dialogue design requires knowledge of the cognitive aspects of the service, of the

environment (type of user, conditions of use), and of the physical interface (keyboard, display).

In another experiment (Marion 1985) a multiservice terminal was simulated that combined both simple telephone services and a number of supplementary services like group directories, electronic mail, automatic voice answering, and calendar management. The terminal could be accessed by a soft key menu. Fifteen students participated in the experiment. The performance time, the efficiency and the number of failures were measured and a questionnaire was submitted to the subjects. Marion found that this system was reasonably easy to use. Performance time decreased significantly with experience. The error rate, however, did not decrease. In general, services consistent (such as using a directory) with prior knowledge of subjects are easier to use than services that are not (such as electronic mail or automatic voice answering).

1.3 Device technology

One of the possibilities to improve the interface of telephone terminals is by using pen computers or automatic speech recognition. In this study a screen based telephone is used that deploys PAID, an interactive display that can be operated with a pen or with a finger. In this section some advantages and disadvantages of pen computers and touch screens are addressed.

Pen computers

Definition

A pen computer is a device which combines a tablet with a flat display and a pen. Users write directly on the screen, which is flat. 'Electronic ink' or markings display the trace of the pen on the screen.

Advantages

Several advantages of using a pen computer are mentioned in the literature:

• Users do not need to develop a new conceptual model of the system because they are familiar with the idea of using pencil and paper (Brocklehurst, 1991 ). For example, the Penpoint operating system uses the concept of a notebook. Penpoint is an object oriented multi-tasking operating system that is especially designed for the needs of

pen-based computers. Penpoint's notebook metaphor provides the concepts of pages and sections. (Carr, 1991)

• Compound actions can be expressed in one single action by pointing and dragging. • The hand-eye coordination is simpler than with a mouse, because with a pen computer

the user can point directly at the space on the screen. (Brocklehurst 1991 ).

• The pen can be used for both writing and drawing so there is no overhead for shifting between these two activities. (Wolf, Rhyne & Briggs, 1992)

• When a character recognition algorithm is used, users can write characters and other symbols (gestures or markings) directly on the screen and have the written input interpreted. (v. Gelderen, 1992)

Using markings has several advantages:

• Different aspects of a mark can be used to control different parameters of one command. (for example, location and shape).

• Commands which need more than one argument can be expressed in one single mark. For example, the mark 'copy' can cover three pieces of information, the command (copy}, what is to be copied (the direct object) and where the copied object is to be placed (the indirect object).

• Markings can eliminate the need for different modes, so there will be no (or less) mode errors. (Kurtenbach & Buxton, 1991 ). In Penpoint for example the location of a gesture controls its intended meaning. Location specific gestures provide an interface that is free of different modes. Throughout this process, the application is in full control of the interpretation and meaning of the pen strokes. (Carr, 1991 ).

• The 'undo' operations also become more flexible, because users do not have to undo their operations in reverse order anymore, they can point out the operation they want to undo. The use of a mark-up editor also fits with version control. For example a new version of a document could be made every time the marks are executed (Goldberg & Goodisman, 1991).

• Gestures could also be used in groupware, for example when different people work on the same document. Groupware can be defined as "the class of applications, for small groups and for organizations, arising from the merging of computers and large

information bases and communications technology" (Ellis, Gibbs & Rein, 1991, p 39). When gestures are used for editing documents, one person can edit the document by marking the text. The other persons can see what corrections the first person made and if they agree they can automatically execute the actual actions (for example, delete a paragraph) (Goldberg & Goodisman, 1991 ).

Gould and Salaun (1987) conducted some experiments to observe how people carry out editing operations with a pencil. The results showed that circles were the most frequently used scoping marks and arrows the most frequently used operators and target indicators. It also showed that the use of gestures has the potential to be faster than a keyboard or a mouse.

Wolf, Rhyne & Briggs (1992) conducted an experiment with We-Met (Window environment-Meeting enhancement tool, a pen based meeting tool). They showed that people found the pen-based interface easy to use, easy to speak to or listen to and write or draw at the same time. This might be important for unskilled typists, because typing tasks ask so much of their cognitive capacity that they can not type and talk in meetings at the same time.

Disadvantages

Using a pen based computer will be most useful and effective if the following conditions are met (v. Gelderen, 1992):

• There should be no delayed feedback of the ink. • The parallax of the screen should be minimized.

• If a character recognizer is used, the recognition algorithm has to be on-line, accurate and quick.

• Errors in the recognition have to be prevented or corrected efficiently and quickly. • When gestures are combined with direct manipulation interfaces, problems can occur

(Aubine, 1992). In most systems a gesture is recognized and the intended command performed after the pen is lifted from the screen. Therefore there is lack of continuous feedback during the interaction. The user can not manipulate the parameters of the command in the presence of application feedback. This is particularly a problem when performing dragging operations in direct manipulation interfaces. Aubine (1992)

developed some methods in which gesture and direct manipulation were combined. In this two-phase interaction technique the recognition takes place during the interaction (for example, when the input device stopped moving or when unambiguous information was available to positively identify the command) and the user can manipulate the parameters interactively.

Pen computers and telephony

When speech and data communication are integrated the use of pen computers is promising. Pen computers could solve some of the problems, mentioned in the previous chapters, users experience when using conventional telephone devices and they can be used to exploit the possibilities of data communication at the same time.

Touch screens

Definition

A touch screen device produces an input signal in response to a touch or movement of the finger on the display (Helander, Moody & Joost, 1988).

Advantages

Greenstein and Arnaut (1988) mention several advantages of using a touch screen: • Touch screens are easy to use and to operate.

• There is direct eye-hand coordination.

• There is a direct relationship between the user's input and the displayed output because the input device is also the output device

Disadvantages

Touch screens have several disadvantages, however (Greenstein & Arnaut, 1988): • The user must sit within arm's read of the display.

• The user's finger may block the view on the screen.

• There is limited target resolution because of the size of the finger of the user. • Parallax can occur when the touch surface is above the target.

Touch screens and telephony

usefulness of touch screens in telephony.

Neumeier (1990) developed two prototypes to investigate the effect of using displays with telephones, a softkey telephone and a touch screen telephone. The softkey telephone consisted of a handset with number keys on the rear. In the center there was a display with on the right side of the display four soft keys for operating the features. On the

extreme right there were eight name keys. An advantage of the softkey telephone is that it is an input mechanism that telephone users know. It only differs from twelve button sets with respect to the large display and the context sensitive functions. In the touch screen telephone the fields of the softkeys are touched directly. The namekeys were replaced by a directory. The advantages of using a touch screen telephone to develop a telephone interface compared to the softkey telephone are:

• The touch screen telephone is very flexible with respect to the number of keys. • It's very useful for integrating a complete phone directory.

• It can easily be turned into an alphanumeric keyboard for typing phone numbers and names (Although this won't be necessary when a handwriting recognizer and a stylus are used).

Bannister, Sheng, Bloedon & Cohn-Sfetcu (1985) mention another benefit of using touch phone: it can guide the user through the logical sequence of phone features. This can be done by using a touch sensitive display that only provides context-sensitive keys and prompts. For example, in the M3000@ touchphone there are approximately 180 softkeys, but less than twelve at one time can be operated by the user. Several personal services are available, such as alphanumeric directories, call logging, and dial-by-name. There is no mechanical dialpad, numbers are dialed by using a virtual keypad on a LCD with a touch-sensitive surface. Bergman, Winlow, Moore, Laidlaw & Carr (1985) compared the dialing performance of this telephone to a standard twelve button telephone. Dialing on the touchphone took ± 10% longer than on the conventional telephone and subjects made three times as many errors with the touchphone. An explanation for this is that the keys used on the touch phone were too small, this was supported by the results of the questionnaire.

A problem with touch phones is the lack of tactile feedback. An experiment was conducted to examine the effect of different combinations of feedback on dialing performance. They tested three kinds of feedback:

• Visual feedback, the key pressed is displayed on the screen. • Clicks, an audible indication that a key is pressed.

• Dual tone multifrequency (DTMF) tones, when a key is pressed on the screen, the same tones are generated in the handset that one would hear when pressing this key on a conventional telephone.

The feedback conditions were: Visual, visual + clicks, visual + DTMF tones and visual + clicks+ DTMF tones. In the last condition the least errors were made. Dialing times, the time interval between dialing the first and the last digit, were not effected by the type of feedback (Bergman et al., 1985).

Comparison of technologies

Roberts and Engelbeck (1989) addressed the effect of device technology on usability of

advanced telephone functions. They prototyped three technologies: 1) a twelve button phone set, 2) a twelve button phone set augmented with speech synthesis, and 3) a twelve button phone set augmented with a bitmapped display and a pointing device (a mouse). The services that were offered to the users included call routing, call screening and message retrieval. With the call routing service the user could redirect an incoming telephone call to a different telephone number. Call screening enabled the user to know who's calling. Message retrieval is the process of listening to any messages that has been left by callers.

The twelve-button set interface consisted of typing mnemonic codes and receiving positive or negative tones as feedback. The set with speech synthesis offered a lot of advantages over tone-alone output:

• The system can prompt the user for applicable input.

• The system can give information that's already in its database

• The system can confirm the correctness of changes by giving its interpretation of the user's actions in natural language.

The interface of the display-based telephone consisted of a display with a keyboard for typing and a mouse for pointing. (They propose to replace the mouse for a touch screen and the keyboard with a handprinting recognizer when this interface would become a real product for the general public.) There are two advantages of a display based interface compared to speech output:

• The information can be scanned when and as long as the user needs it.

• It is not necessary to memorize commands because the parameters can be changed by direct manipulation instead of by issuing commands.

Roberts and Engelbeck conducted an experiment to investigate whether there are significant empirical differences between the three technologies. The time to complete a task, the amount of errors, the learning effect and the subjective preferences were measured. The participants varied in age, education and profession. With the screen phone it took the least time to complete a task, but there were no significant differences between the error rates of the systems. The screen phone was preferred over the other systems. There were learning and forgetting effects, but they didn't differ significantly across the systems. A problem that users had with all systems is translating the statement of certain complicated tasks into system parameters that needed to be changed. Future systems design should match systems' alternatives more closely to likely tasks.

ltoh, Koike & Yoshida (1988) examined the efficiency of name access. They used two methods: voice dialing and touch dialing. Name access frees the user of telephone numbers, because they can place a call directly from the directory. Each user had two directories: one made by and for the user and the other was a database. When a name from the ready made directory was called for the first time, the name and telephone number were copied to the personal directory. The following results were found: • Users felt that name access is superior to conventional speed dialing.

• The touch phone became more convenient as telephone use increased.

• In both name access systems, more than half of all the calls were made by name accessing from the personalized directory.

1.4 Conclusions

The literature suggests that using a pen computer or a touch screen as an interface for the telephone could solve usability problems that users experience when operating standard twelve button telephone sets. A screen-based interface could also be useful to combine telephone features with other tasks, such as calendar management, making memo's and interactive note exchange. The addition of these functions could be very important to support the activities of the secretaries. The interface should support multi-tasking by the use of reminders and by minimizing the memory load the telephone lays upon its users. The features should be easy to learn and easy to use. In this way the interface could make interruptions less disruptive, make it easier to make notes during telephone calls, and to leave messages to absent people.

The evaluation of such a telephone should take into account the context in which the system will be used. Tests have to be done with future users and in real situations in order to be able to determine the use and acceptability of communication systems. In the next section some issues concerning the evaluation of the prototype, the prototype, and the objective of the experiment will be discussed.

1.5 The experiment

PhoneO

The design of the interface of the first prototype of the telephone (PhoneO) was based on the button concept of standard telephone sets. The prototype of this telephone provided the following functions:

• Calling. Calls could be made by using a virtual keypad on the screen or by using the hardkeys on the telephone set.

• Making and sending notes (off-line). The sending of the notes was simulated. • Making a drawing in connection with the person you are calling (on-line).

• A list of names and telephone numbers. Names and numbers could be entered with a virtual keyboard that popped up on the screen. Calls can be made by selecting a name and pushing the 'dial' button.

Features like call transfer or conference call were not implemented yet. Evaluation methods

There are several methods that can be used to evaluate user interfaces. Yamagishi &

Azuma (1987) compared four evaluation tools with respect to isolation of specific problem areas, the difference between new and old versions and the difference between designers evaluation and non-designer evaluation. These tools were: protocol analysis, interview, evaluation questionnaire and logging data analysis.

The results per method were:

• Protocol analysis: Problem areas were extracted from the recorded videotape in a problem report. This method was most time consuming.

• Interview: Critiques obtained in a interview were summarized in a sheet. The results of the interviews were consistent with the results of the protocol analysis. These

retrospective reports addressed more global and higher level issues and were useful to analyze the semantics of quantitative evaluation. This method was also rather time

consuming.

• Evaluation form: This was useful for comparisons between versions of the interface and between designers and non-designers. It produced quantitative results, but contained no information about what and how improvements specifically should be implemented. • Automated data collection: This method produced quantitative results but was unable to

be used to indicate users satisfaction.

Root & Draper (1983) evaluated the use of questionnaires and found that check-list style questions about specific existing features of a system do yield findings that are robust across methods of administration and across levels of user experience. The method is not useful to identify omissions; to find omissions open questions should be asked.

The different tools can be used in laboratory studies or in field studies. An advantage of a laboratory study is that they can be conducted fast and that they are relatively cheap. The problem with this kind of studies is that they exclude a lot of contextual variables that determine the acceptance and use of the product. Field studies on the other hand are conducted in a more realistic context. Disadvantages of this kind of studies are that they are relatively expensive and require 'close to production' quality from hardware and software. Another problem with field studies is that there are little possibilities for experimental manipulations. In the MICE (modular integrated communications

environment) project the researchers tried to provide means for research on the design and evaluation of advanced communication services and user interface technologies in a realistic environment, but without the expenses and the overhead of field trials. It is a real system, tested with their colleagues as subjects. It permitted experimental control over services. During system failure, MICE users are automatically connected back to the telephone company for service. (Herman, Ordun & Riley, 1986).

The design of communication technology differs from the design of office automation and single user applications because in communication systems the social element is an integral property of the system, not an accidental one. From the user's point of view, the capabilities provided by the system, the rules for its use and its reactions to their actions depend jointly on what is implemented and how other users behave.

There are four factors that designers have to take into account when they are developing a communications system (Cool, Fish, Kraut, & Lowery, 1992):

• System drift. System behavior changes as social norms change; communication systems can change while technology does not.

• User concern about the perceptions of others, in which individuals shape their use of the technology based on their beliefs about how others will evaluate them. People are concerned with implicit communication as wit explicit communication.

• Conflicting social goals, in which the different roles that the same user plays may impede the achievement of group goals.

• Critical mass, in which the requirement to have a sufficient number of people using a system places constraints on the rapidity of iteration.

While the basic cycle of build-study-redesign is still valid, the implementation should be more complete and robust and the study should extend over a longer time period before one trusts conclusions about the ultimate value of the communications systems. Changes need to be evaluated in the context of system use, rather than in the laboratory.

Objective

According to Herring (1993) a software simulation can quickly predict usability problems of later hardware designs. Behavioral records combined with thinking-aloud protocols could provide useful information at an early stage of the design process, in a cost-effective manner (Wright & Monk, 1991 ).

Because PhoneO didn't have 'close to production' quality, a laboratory experiment was conducted to gather base-line information about the initial acceptance and usability of the

PAID telephone.

The following issues were addressed:

• Do users understand the concept of a graphical interface on the telephone terminal? Although people become more familiar with graphical user interfaces, telephone

interfaces usually do not have this characteristic. It has to be investigated whether such an interface is fit for telephone terminals.

• Do users understand the functionality of the telephone?

With the arrival of ISDN additional services will become available to telephone users. In PhoneO these services consist of making and sending (off-line or on-line) notes. The interface should make it easy for users to understand these services, to make sure that they can and will be used.

• How do users interact with the telephone?

In PhoneO there are several ways to make a call and it is possible to make notes with the system. The way they make use of these features have to investigated, in order to make the telephone as efficient and effective as possible.

• Does the graphical interface support multi-tasking of users?

Secretaries are often engaged in multi-tasking and are often interrupted. Therefore the interface has to be designed in a way that makes it easy to suspend an activity and to resume these suspended activities.

Chapter 2

Method

2.1 Subjects

Five male and five female subjects, all aged between 18 and 40 participated in the

experiment. Seven subjects were paid and three were not. All subjects were right handed. They all had some experience with wordprocessing, but no experience with the prototype of the telephone. All subjects were students except subject three (secretary) and subject seven (nurse). The demographic data are shown in table 2.1

Subject d'/'Ji.

Education

computer use telephone use

1 d' HTS > 1 a day (H & W) Command 1-6 times a week (P)

2 d' TU 1-6 times a week Graphical 1-5 times a day

(H & W) (P)

3 ~ Mulo-A > 1 a day (W) Command > 5 times a day

(8)

4 ~ HEAO 1-6 times a week Command > 5 times a day

(W) (P)

5 ~ TU 1-6 times a week Command 1-5 times a day (B

(H & W) & P)

6 d' TU > 1 a day (W) Graphical 1-5 times a day (P)

7 ~ HBO-V < 1 a day (H) Command 1-5 times a day (P)

8 d' TU 1-6 times a day Command 1-5 times a day

(W) (P)

9 ~

vwo

10 d' TU > 1 a day (H & W) Graphical 1-5 times a day (P)

Table 2. 1 Demographic data of the subjects participating in the experiment. VWO and MULO-A are at highschool level, the rest is equivalent to college-level. The command based systems consisted of MS/DOS or UNIX, the graphical interfaces consisted of

MS/WINDOWS, X-WINDOWS or MacIntosh. H

=

=

2.2 Material

The sessions were recorded on a Philips audio recorder and a Sony video recorder. Three telephone modules were used, one by the subjects and the other two by the experimenter. The telephone modules consisted of three parts: A Siemens Euroset 821 telephone, PAID and a 486 Dell PC with an ISDN card. The PAID consisted of a 9.4" back-illuminated LCD with 480 rows of 640 pixels. 480 rows of 320 pixels were used for the interface. The rest of the screen was covered with removable flaps. The stylus was corded. The PAID was linked to the PC. The three telephone modules were linked together by an ISDN-switch (See figure 2.1 ). ISDN --"---jswitch Analog telephone Pc + Application program+ PAID.

Figure 2. 1 Configuration of the telephone

systems.

The subjects only interacted with the PAID and the telephone, the PC was hidden. The buttons on the PAID they pressed were logged with the system time attached to it. In task 7-12 the subjects used a Philips XT Pc with WordPerfect 5.1 for the typing tasks.

The interface consisted of three main screens: the keypad screen, the names screen and the notes screen:

• The keypad screen (See figure 2.2). In this screen users can call a person by pushing the buttons on the virtual keypad. The digits appear in the white bar above the virtual keypad. It is possible to correct a digit by pushing the backspace button next to the white bar, or to clear the entire telephone number by pushing the "clear" button. Dialing of the telephone number is invoked by pushing the "dial" button. The status information

appears at the top of the screen. Guidance about the actions a user should take (for example, "pick up the handset") appear as message boxes in the middle of the screen. From the keypad screen the user can go to the names screen or the notes screen. • The names screen (See figure 2.3.a). In this screen users can call someone by selecting

the name of the person they want to call. Dialing of the corresponding telephone number is invoked by pushing the "dial" button. The status information and information about actions users should take are displayed in the same way as in the keypad screen. Users can select a page by pushing the tabs next to the page. To enter a new name, the "input" button has to be pushed. If this button is pushed a virtual keyboard appears (See figure 2.3.b). A page can be selected by pushing the arrows next to the names. The name and number are entered by pushing the letters and digits on the virtual keyboard. If the name and number are entered, the "put" button has to be pushed to insert the

Keypad

I

IG

OJ

0

[TI

EJ0~

_I

I

0 0 ~

[] ~0

-

~ypad

I

I I

I

E

Figure 2.2 Schematic picture of the keypad screen Names Pietersen 080-12345 Pyters 345 Pyz 08370-21121 ABC COE Names/Input Pietersen 080-12345

ct=

~

~

I

11

11

I B

name in the list. If subjects want to edit a name, they have to select the name, and push the "get" button to get the name out of the list. Then they can edit or clear the name. It is also possible to call from this screen. Users can go back to the names screen by

pushing the "names" button on the bottom of the screen. From the names screen users can go to the keypad screen or the notes screen.

• The notes screen (See figure 2.4.a). In this screen users can make notes for themselves by drawing with the pen on the screen. When they are calling with someone they can draw together. The notes are automatically saved. Notes can be selected by pushing the tabs next to the note. It is also possible to send a note off-line. This is done by pushing the "to" button. When this button is pushed, a virtual keyboard appears (See figure 2.4.b). The user enters the telephone number and can also enter the name of the

addressee. After the number is entered the name and number can be put above the note by pushing the "put" button. The mail is sent by pushing the "mail" button. To get back to the notes screen the "notes" button has to be pushed. From the notes screen users can go to the keypad screen or the names screen.

Notes 1 2 3 I Mail

~

~

[;]

t•ypad

I I

Ncrn~

11

I B

EJ

Figure 2.4a Schematic picture of the notes Figure 2.4b Schematic picture of the mail

screen. screen.

2.3 Setting

The experiment was conducted at IPO. The subject was in one room, the two

experimenters in another (see figure 2.5). The experimenters could watch the subject on the monitor. The screen of the subject was linked to another screen, so the experimenters could see which part of the task the subject was executing. This was important for the

timing of the interruption. The experimenter could also listen to the subject during the experiment through a speaker.

Room1 Phone, nr 10 Wicrophon Room2

_r:::J

c::J

~

Figure 2.5 Setting of the experiment.

2.4 Procedure

Before the experiment started, the directory and the notes were cleared. The subjects had the telephone and the pc on their desk. They were allowed to use pencil and paper when they had to fill in a form or a questionnaire, not during the execution of the twelve tasks. First the subjects filled in a form (see appendix A) with background information about their computer experience, frequency of telephone use etc., then the subjects that were paid filled in a declaration form.

The subjects were given the first page of the instruction, (see appendix B) and after that they were given a demonstration of all the features of the telephone.

Subjects who did not have any experience with thinking aloud during an experiment were given an exercise to practice this skill.

The subjects were given the instruction for tasks 1-6 (see appendix B) and executed them. After these six tasks there was a coffee break. During this break the subjects were given a refreshment and the telephones of the subject and the experimenters were restarted. This

was done to minimize the chance of a system failure2 •

After the break, the subjects were given the instruction for task 7-12 (see appendix B) and they executed them.

They filled in a questionnaire concerning the interface. (see appendix C).

The subjects were supposed to complete all tasks, but the experimenter interfered when a task was too difficult or when there was a system failure. When a system failure occurred the subject was interrupted, the system restarted, and the subject was asked to start the current task again. When a task was too difficult, the subject was interrupted and the experimenter explained to the subject how to proceed.

The subjects also had access to a manual. They were asked only to use it when they could not proceed without it.

2.5 Tasks

Each subject was given twelve tasks, in task 1-6 only the telephone was involved ('isolated' use condition), in task 7-12 wordprocessing and telephony were involved to simulate a daily use situation ('embedded' use condition).

Because the pattern of activities of secretaries is governed by multi-tasking, scheduling, and lack of control the subjects unexpectedly were given extra tasks and were interrupted in both the conditions.

A pilot study was conducted to test the suitability of the tasks and conditions and to determine the time needed to complete the experiment. It was found that the tasks were relatively hard. It was expected that it would be too difficult for the subjects to interrupt them at the start of the experiment, therefore all the subjects performed the tasks in the same order.

The subjects were interrupted at a certain point in the task and not after a certain time. This was done to make sure that the interruption occurred during the execution phase (Norman, 1986) of the task and not between to phases (for example, between execution and evaluation) or after the evaluation of the task.

Table 2.2 shows the assignment of the tasks to the conditions.

isolated use embedded use

dedicated task 1,2 task 7,8

mu~tasking task 3,4 task 9,10

interrupted task 5,6 task 11, 12

Table 2.2. Assignment of the tasks to the conditions.

The tasks are based on the top ten list of activities of secretaries found in the study of Brouwer-Janse et al. (1992). These activities are:

1. Answering, making, and transferring phone calls. 2. Being interrupted by phone calls and by persons. 3. Address file and archive management.

2_{Because the experiment with the prototype was done very early in the development}