4. System Design



The computer package documented in this thesis has undergone a number of changes since the original design, both as a result of new ideas, and as a response to user-testing. The following section looks at some general features of the package, at details of the various modules of the package, and briefly mentions, some of the changes that have taken place. Full details of changes to the package which have been implemented can be found in chapter 6, and ideas for proposed changes in chapter 7.

Initially, the idea was to produce an easy to use computer system which would allow children to describe people, places, and feelings. Most of the psychologists on the team had little or no idea about what was possible within a Macintosh computer program, and so a lot of time was spent discussing different interface possibilities, and ways of asking questions that would allow the child sufficient scope to be able to use the system, without getting lost in its intricacies. The approach that was adopted was the production of a number of different program modules. The interface to these could be designed by the whole team, coded by the programmer, and tested on children. Modifications could then be made if appropriate. From the start the intention was to utilise the full graphical and audio facilities of the Macintosh to produce something which a child would find fun to use. In order to do this it was necessary to consider how the system would interact with the child.


4.1 General Interface Items



4.1.1 Textual Input

In the initial stages it was hoped that we could produce a package which needed no textual input whatsoever. It had been assumed that a keyboard would be an obstacle to communicating with the child. It was discovered on the first tests of the package with children in the target age-range, that a keyboard is something that they have all met, and know how to use. It was decided that it was practical to use the keyboard for a number of tasks, and in order to make the keyboard less difficult to use its response was simplified in a number of ways:

1) Key repeat is turned off, so keys do not accidentally repeat if a user holds down a key for too long. Users who are unfamiliar with computer keyboards tend to be put off by keys which repeat too quickly. Additionally, the perception of "too quickly" varies from user to user.

2) Only a particular set of characters are permitted - preventing the entry of erroneous characters, or unwanted symbols.

3) A button, marked "oops", is provided next to the text-input box. When clicked, it will delete everything that has been typed into the current text box so that users do not need to understand the concept of the backspace key. Since key-repeat is disabled it is easier to completely delete a piece of text that was typed in error with the oops button, than having to press the backspace key repeatedly. However, the backspace key was left operational since it was discovered that most of the target group knew about its use.

These modifications to the Macintosh interface were made to simplify the interface for young children, whilst not being off-putting to older children and adults. One valuable aspect of asking the child to type information, is that it tells the interviewer about the abilities of the child. It may show that a child is going to have spelling difficulties (though this may be obvious only if the child types more than their first name etc.).

4.1.2 Messages to Users

In order to make the interface as user-friendly as possible, a decision was made to incorporate an animated animal character who could announce messages to the user, rather than the typical computer text-based message system. Text-based systems are able to present messages to users in a form which is readily understandable providing sufficient thought is given to framing the messages appropriately. However, it was thought that a package employing such an interface would be less likely to succeed, as a wholly text-based system would be less likely to capture the attention of children. The initial form the "friendly" interface took was of an animated talking Moose, whose speech was produced using a standard Macintosh phoneme-based voice synthesiser called "MacinTalk". Sentences were constructed using phonemes and these were incorporated into early versions of the package. The phoneme-based system gave the power to create a vocabulary of great complexity whilst saving on storage requirements, but at the cost of decreased clarity. This system was tested on subject groups of children within the age range for the package. It was discovered during tests on small groups of children, that they initially had great difficulty understanding such speech, especially if the sound generated was not amplified by an external device, but that most of them would eventually pick up the "accent" of the synthesiser. A further problem was that such speech could at best provide words that a listener could only guess at, particularly some of the basic speech sounds, such as "ing", which are notoriously difficult to reproduce on a phoneme-based synthesiser. Words that contained these particular phonemes had to be avoided. For these reasons it was decided the MacinTalk system for generating speech was not good enough, and so the search for another speech system began.

The next step was to use a system of individual words, put together in real time to produce the desired sentences. This system utilised a bird character, animated by showing consecutively one of a set of three bird animation frames (figure 4.1), with the bird's beak in different stages of opening, apparently synchronised to the words spoken. A female voice was used, as it was thought likely to be less threatening to the child. This speech system gave greater clarity, but at a cost of increased storage of more than 800 times as much for a few sentences compared with the phoneme-based approach. Another problem with the stored-word system is the fixed inflection in the individual words, giving a very stilted sound. Whilst not an ideal system, it was an improvement on the previous system and was adequate for demonstration and test purposes.


fig. 4.1 - three frames of the different positions of the talking bird


The use of such a speech system was designed in such a way as to enable the simple addition of extra words and phrases, and even the use of alternate languages (see Appendix E). All the words and sentences used by the package for particular situations are stored in a separate file. The one which the package normally uses is called "Standard", and the file to be used can be selected when the package starts up by clicking on an alternate file of words and sentences. Any other voice files will add to the storage requirements of the package as a whole, but it is envisaged that anyone using the package would normally have a single voice file to use with the package, rather than many voice files, taking up large amounts of disk space. This system has been modified to permit the bird to display its spoken text as written text in a speech balloon (see figures 4.2 and 4.3). This gives both a visual back-up to the sound, and could act as an aid to those who may have hearing difficulties.

fig. 4.2 - bird animation frame complete with speech balloon.


fig. 4.3 - the three bird frames complete with part of the speech bubble.


Looking ahead to the next stage of speech support, and where the operating system of the host Macintosh is of a version that supports the compression software required, a technique will be employed that utilises complete phrases, stored as compressed sounds. Two compression ratios are available in Macintosh system software version 6.07 and above, these being 3:1 and 6:1. Empirical tests will be performed to discover the optimum compression ratio. This new system will give sentences of great clarity if the individual phrases are not compressed too much. A system was added to the package, which allowed the operator of the package to turn off the bird animated character whilst maintaining the spoken words. It was felt that in some cases, the animated character might be inappropriate, so an alternate character can be used. If the speech file being used contains an appropriate sequence of pictures then these will be used instead of the standard bird character. These options allows the operator of the package to choose the style of an important part of the "friendly interface".

A further possibility for future expansion is the use of a recent addition to the Macintosh Operating System, a feature called "QuickTime". This permits the storage of full-motion video, complete with sound. QuickTime uses a compression system to reduce the storage requirements of video and sound, which would otherwise be excessive. QuickTime will permit the package to have video sequences of a real person speaking, which will be of particular importance if, for example, the person speaking also uses sign-language. The quality of the video reproduction of QuickTime, should be sufficient for sign-language, and possibly also for lip-reading. QuickTime is available to any colour-capable Macintosh computer.

4.1.3 The Sticky Finger

A number of individuals who had never used a Macintosh before were asked to use the mouse to operate a program. From these simple tests it was determined that some people without "mouse expertise" found it difficult to hold down the mouse button while moving the mouse (known as "dragging" the mouse), which is a standard operation for the Macintosh user interface. Many people experience this difficulty, and some of the test computers used a "track ball" as their mouse, which was even more difficult for some people to move while holding down the "mouse"-button. To overcome these problems the concept of the "sticky finger" was created, where items that may be moved with the mouse, stick to it when the mouse is clicked on them, with the cursor then changing to a grabbing hand (figure 4.4). Any item which will stick when it is clicked upon causes the mouse cursor to change into a hand with a pointing finger (the "pointing hand") whenever the mouse cursor moves near or over them. All those parts of the package which need things to be dragged around utilise this effect in an attempt to provide consistency across the various modules.


fig. 4.4 - the "grabbing hand" and "pointing hand" cursors



4.2 Modular Design



The package consists of a number of modules:

4.2.1 Introduction
This module presents the child with a picture of a boy and girl from which they can select their gender. The child is then prompted to enter their name and age. The gender information is used to set the gender of material presented in other sections of the package. The answers to these questions can be verified with known information about the child.

4.2.2 Emotions I
In this module a view of an individual of the same gender as the child is presented. The child may select an expression from a palette of seven emotional expressions, and talk about them. This allows the interviewer to discover if the child understands the emotions represented in the palette, and to ask the child to think about events that may have a specific emotional aspect. It is designed to be used in whatever way the interviewer thinks is appropriate, and information gleaned from its use can be used to tailor the way in which the interviewer presents the other modules - for example information about the child's use of language.

4.2.3 Emotions II - similar to Emotions I except, by default, eleven different scenes are presented featuring an individual of the same gender as the child (see appendix A for pictures of the scenes). The child chooses an expression from the palette that they believe is appropriate to the scene in question. The standard set of scenes allow the interviewer to explore the child's understanding of the emotions represented in the emotional palette. This module can be used with a set of alternative scenes, selected by the interviewer from a collection of many different scenes, with specific features.

4.2.4 Buildings - Used to select a building or place from a range of alternatives. The information gathered by this module about locations is likely to be verifiable, and can be used as an indicator that the child is using the package reliably. This module can be used to permit the child to choose a building to represent any place that the interviewer wishes them to talk about.

4.2.5 People - Permits the selection of one or more individuals (usually associated with a place). This is used initially to get the child to describe the people who live in their primary residence. By populating a location the child can be asked questions about how they feel with individuals when they are in certain locations.

4.2.6 Emotions & People - For each of the seven emotional expressions in the palette, the child is able to select people that they have felt that emotion with. The people that the child chooses are ones who they have previously associated with a place in the buildings module.

4.2.7 Somatic Experiences - Allows the marking of pain sites on pictures of the front and back views of a child. Pain sites may then have a shape, size, and "throb" associated with them. This module is not based on any existing aspect of the SAGE system.

4.2.8 Environment - Used to place items of furniture into a "room". This module was initially thought of as an adjunct to the buildings module.

These modules have been designed and produced at various stages in the package development, and there are plans for some further modules to be introduced in the future. These are discussed in chapter 7.


4.3 Customising The Package



A number of the modules of the package can be easily customised in order to modify the way in which they are presented to a particular user. This is possible because when certain modules initialise, they check to see what resources are available to them, and are able to modify their behaviour if certain alternate resources are found. As described in appendix B.5, Macintosh files consist of two parts or forks, and one of these holds the resources that the package uses. (Resources are discussed in appendix B.6). More details on the resources seen in typical Macintosh programs can be found in Appendix E and Appendix F. This ability for simple customisation is designed to allow the package a degree of flexibility, and it is envisaged that tools will be provided to users to permit their customisation of the package, where this is within guidelines defined by the psychologists on the team.


4.4 "Choosing" Tools



A variety of tools are used in different parts of the package to allow the user to choose from a number of available options. It is important, when looking at the design of such tools, to bear in mind three key questions [Vertelney et al, 1990]:

We will look at each of the tools in the various modules in turn.

4.4.1 The Emotions Tool

The user is presented with a scene containing a child with a blank face, and a rectangle comprising a set of seven equal-sized rectangles arranged horizontally, each of which contains a different emotional expression (known as the emotional palette, or emotions tool - see figure 4.5). Clicking on one of the rectangles transfers the expression within to the face of the child in the scene, and puts an indicator, a coloured rectangle, around the chosen expression in the tool. Once an expression has been chosen, though it may be replaced by a different expression, it is not possible to subsequently have no expression. The tool has no facility for removal - only replacing, since the tool is used when a selection is mandatory. The emotions tool stays in place after an expression is chosen and acts as a visual reference, so that the child can compare the one they have chosen with the remaining expressions. Figure 4.6 shows two views of a scene, the first without an expression, and the second after the frightened expression has been added. It is important to note that the user may go on to change the expression on the second view to a different one simply by selecting an alternate expression from the tool.



fig. 4.5 - the emotions tool


The seven different stylised emotions represented here, were intended to give a sufficient breadth of expression. They have been designated "very happy", "happy", "neutral", "sad", "angry", "ambivalent", and "scared". These expressions evolved from designs which were based upon stylised emotions as represented on the faces of characters in children's comics. At the outset children may not recognise such stylised emotions, and only come to know them through the use of them in cartoons. These expressions were tested on groups of children in the target age, to see if they assigned to them the same "feelings" that we had designed them to represent. More details on the test results are given in Chapter 5, and modifications made as a result are documented in chapter 6. Following adjustments to the designs and further tests, it was possible to accept the expressions, and to be able to say that children assigned the same sorts of feelings to them, with minimal deviation, as those that we had originally intended to convey.

fig. 4.6 - example of a scene without expression and with the "new" frightened expression.



4.4.2 The Scrapbook - A Rejected Tool

Originally a scrapbook (figure 4.7) was used to present a series of pictures of items, where the intention was to get the user to choose just one of them. The scrapbook was based on the Notepad and Scrapbook programs for the Macintosh. These two programs have been part of the Macintosh System software since the earliest Macintosh computers. This meant that they were well known to anyone who had used a Macintosh. For anyone who had used a Macintosh, which included the whole of the project team, use of the Scrapbook was intuitive, however, this obscured the fact that our target audience of children found it too difficult to use. In the initial design for the Scrapbook browser, the user looked through the scrapbook, going forwards or backwards by clicking in the "dog-eared" corner of the page, and then when they decided on the item that they wanted they clicked on the item, whereupon the page containing that item was torn out of the scrapbook and presented at the side of the book. The scrapbook itself was closed. If the user decided that they would like to look again they could go back to browsing the scrapbook. If they chose a different item, then that was torn out of the scrapbook, and the initial choice returned to the book, where the process could start again. When the user was finally happy with the choice that they had made, they clicked on the "OK" button presented elsewhere on the screen, in the area which was set aside for such buttons. This initial Scrapbook design was refined in the light of user-testing and a slightly different scheme was employed. Since the user had already met, and would meet again later, the idea of dragging objects, the way that an item was chosen from the Scrapbook was modified, and the new system was to be that of dragging the item from the page of the Scrapbook to a pin-board, on to which the dragged picture was placed. If a different picture was dragged there, the initial picture was returned to the Scrapbook. The picture on the pin-board, could also be dragged back onto the Scrapbook, in order to deselect it.



fig. 4.7 - the two different views of the scrapbook - closed and open.


The scrapbook tool is theoretically able to utilise any drawing as a representation of a dwelling. In practice it is intended that the person supervising the running of the package will have selected an appropriate set of pictures for the particular child who is to use it. In this way it becomes simple to target the package for a particular user. This is done is by the use of a custom resource type. Resources are discussed in appendix B.6 and section 4.3, with further details on the definition of resources in Appendix E and resource types in Appendix F.

The "Bild"resource is a custom resource, designed to be used by this package. It contains information relating to the picture of the building. It includes, for example, the rectangle which defines the door to the house, enabling the package to place the number of the house on its door! There is one resource for each house picture. All pictures corresponding to such resources are placed into the scrapbook for display.

This tool was initially tested on a small group of children, who were able, with some difficulty, to operate it and select from it a picture of the house that they lived in. A further problem was encountered with use of the Scrapbook for the selection of houses, in that the pictures of the different kinds of houses were not specific enough to be recognised by the user. It would be difficult and unwieldy to represent all the different possible types of dwellings, and so the mechanism of the Scrapbook for use for choosing a dwelling will be replaced by another system, possibly one that presents the user with basic building blocks, and then allows them to put together a more representative image of their own house by dragging extra items onto the picture.

4.4.3 Family Chooser

The user is presented with a "global family" from which individuals, and other families can be created.

fig. 4.8 - the "global" family


This family was created to represent a reasonable set of alternatives for height and/or age (figure 4.8). It can be customised for a particular target user, to include particular types of individual. The users click he member of the family that they wish to select. The coordinates occupied by each member of the global family are compared with the coordinates of the mouse-click. If the mouse-click position lies within the coordinates occupied by a global-family member then this family member is selected. As each person is selected, they are named by the child in response to a dialog box. This is so that they may be referred to later, and as a reminder of who has been chosen (and hence who remains to be chosen), since the same representation may be used to depict a number of different people. As they are selected, a reminder of the chosen set are shown in a small window to the top-left of the main window. This acts as a visual reminder to the child of who has been chosen.

In the original version of this tool, the locations of the members of the family were hard-coded, their values were written into an array from constants within the source code of the package. This meant that in order to adjust the contents of the global family, for example if we wished to add ethnic characters, the source code of the package would have to be modified because the positions of all the family members would be different from the original values. This was deemed to be too limiting a system. The original module had the ability to display a family that the user chose from the global family, up to a limit of 25 persons. As the number of persons chosen made the new family larger than would fit on the screen, the new family was drawn at a reduced size, whilst maintaining the height-to-width ratio of the family members. The routine which drew the new family calculated a set of coordinates for each family member, in order to determine when the new family became too wide to fit the screen. It was a logical extension to use this family drawing routine to draw the global family as well, and by returning the coordinate information to the calling routine the information could be used by that part of the program that watches for a mouse-click on the global family members. This technique had three distinct advantages:

1. It meant that the global family could be customised without any changes to the source code of the package at all. There is a resource of type "Fmly" for each person in the global family. When the package starts up each "Fmly" resource is found, and the picture associated with each "Fmly" resource is collected, so that it may represent this individual in the global family.

2. The routine which drew the family could be reused in another part of the package, reducing the amount of programming code required.

3. Modifications to the way that families were drawn in all modules could be achieved by changes to a single drawing routine.

With hindsight it is obvious that this should have been the direction taken from the outset, but at the time, simplicity of coding had been the intention, and the routine which drew the selected family had not been written at that time.
Any sort of picture may be used to represent a "family member", including animals such as cats and dogs. This feature has been exploited in the package, to allow adaptations to the original global family in the light of responses from children who have used the package. More details of the changes can be found in section 6.6. It is intended that the package as supplied to psychologists will come with a set of artwork of characters of different ethnic origins, and that customisation of the global family will be customised using a program module designed to allow the psychologist to target the operation of the package more specifically.

4.4.4 Emotions and People

This particular module of the package reuses an existing tool as an indicator (figure 4.9), and introduces a new tool. The Emotions tool is used to indicate feelings, so as to facilitate discussion of the child's experience of those feelings. Each of the seven basic emotions is presented, along with the question "Have you ever felt like this?". The emotions themselves are presented in a sequence where "negative" emotions are interspersed with "positive" ones. The emotions tool ranges the emotions from positive to negative from left to right. A dialog box with "Yes" and "No" is also presented, allowing the user to give their answer.




fig. 4.9 - Emotions tool being used as an indicator.



If the user's answer to the question is "No", then the next emotion in the sequence is presented. If they answer "Yes", then a short animated sequence takes place, transferring the expression of the emotion to the head and shoulders picture in the left-hand-corner of the window, while all the other emotional expressions fade out - so that only the current one remains (figure 4.10).


fig. 4.10 - User answers "Yes" to question "Have you ever felt like this?"


After all the emotions have been dealt with a new tool is presented (figure 4.11). This tool has three parts:

1. The top half of the display shows the family chosen by the user in the family module.

2. Bottom left is a reminder of the emotion that is being investigated.

3. Most of the bottom half of the display has space for the user, and can also hold members of the family that they choose.


fig. 4.11 - Emotions and People main tool.


The user is asked about each emotion in turn, and may click on a member of the family with whom they felt that particular emotion. Upon selecting a family member, the person is moved to the bottom part of the display to stand beside them, and a Yes/No dialog box appears to make sure that they really meant to select that person (figure 4.12).


fig. 4.12 - the user chooses a family member.



If the user clicks "No", then the family member returns to the family section of the display and returns to its greyed out state. If they click "Yes", then the family member returns to the family section, but is drawn in black to highlight it (figure 4.13).


fig. 4.13 - the selected individual returns to the family section of the tool.


The interface documented here is the third implementation of this tool. Details of the previous versions can be found in section 6.8, and details of a proposal for a different interface can be found in chapter 7.

4.4.5 Somatic Experiences

Obviously, somatic experiences of most interest to an interviewer consist of pain and/or discomfort. This tool (figure 4.14) allows the user to mark pain sites on a representation of themselves, using pain-shapes, some of which have an animated representation. The tool has the facility for representing three aspects of pain: location, shape, and size/intensity.


fig. 4.14 Somatic experiences tool. On the left the user has selected a pain-site by leaving a gremlin mark. On the right the user has clicked the pain palette and a representation for that pain has replaced the gremlin. The user may now adjust the size and throb-speed of the pain-spot.


The user is asked about their history of pain, starting with the idea of pain experienced in the past, and then continuing on to pain currently being experienced, or the last time the user had pain. The user is presented with the front and back views of a child of the appropriate gender, and then asked to mark an area where they had pain, by clicking which leaves a gremlin mark. They are then able to choose from a palette of different types of pain representation. The palette contains a number of pictures of pain types. The pictures themselves may be a sequence of pictures, i.e. an animated representation, representing a pain type; for example a flashing bolt of lightning. Upon selecting a pain type, a standard-sized representation of it will transfer to the site of the pain marked by the user. They then have the facility for choosing one of five different sizes/intensities for the pain spot. The final stage is to set a throb-speed for the pain spot. This is achieved by manipulating the slider control upwards to increase, or downwards to decrease the rate of throb.

There are three main parts to the somatic experiences tool (figure 4.15).

The pain-type palette. This appears on the right-hand-side of the somatic experiences tool window and is a dynamic item: it re-sizes depending on the number of different pain-type representations that are available when the tool is initialised. The standard set of pain types in the tool were based on children's drawings of pain or hurt.

The size/intensity palette. This was designed to offer a range of alternative sizes without making the manipulation of size a complicated matter.

The throb-speed controller. This uses a standard Macintosh interface item: a slider control, usually found on scrolling windows.


fig. 4.15 The three main parts to the somatic experiences tool. Pain-type palette, throb-slider and size/intensity palette.


Changes which were made to the somatic experiences module over the course of its development are detailed in section 6.9, and results from trials using it are presented in chapter 5.

4.4.6 The Slide-Bar



fig. 4.16 - the slide-bar (shown with all four slide positions empty).


The slide-bar is a tool which permits the choosing of an item from a selection of items, and the subsequent placing of the chosen item at a location in the main window. The slide-bar (figure 4.16) displays up to four items at once from which a selection can be made. It can display more than four items by utilising the arrow buttons at both ends of the bar to scroll through the set of "slides" on display. Selection of items from the slide-bar is made by clicking on a slide. This causes four things to happen (figure 4.17):

1. A rising note sound is played to indicate something being picked up.

2. A full-size copy of the selected item appears near to the cursor.

3. The cursor becomes the "sticky finger" (see 4.1.3).

4. The slide bar becomes black, obscuring all the other slides.

As the user moves the mouse (figure 4.18) the selected item follows it (see appendix G for details of the animation algorithm used). To place the item in the main window the user clicks the mouse again. This releases the item from the sticky-finger (figure 4.19), and plays a falling note sound to suggest something being dropped. If the user had clicked the black slide-bar area while the item was attached to the cursor, the item would have returned to the slide bar, as though it had never been selected, and no sound would have played.


fig. 4.17 The mouse has been clicked over the teddy-bear slide, and a teddy-bear picture has appeared near to the cursor in the main area of the display, while the slide-bar itself has turned black.



fig. 4.18. The teddy-bear picture is connected to the sticky-finger cursor, and follows it as it is moved by the user.




fig. 4.19 The mouse has been clicked, releasing the teddy-bear picture. The slide-bar is redrawn to show that other items may now be selected.



The slide bar is an attempt to present the user with an interface in which it is easy to select from a number of alternatives, but where other items are in view in order to have a visual reference to them. Note that where the choice is between similar items, it would be appropriate to group them together. The visual display of the slide-bar acts as an aid to memory. Section 7.1.1 discusses a possible alternative use for the slide bar.

Unlike the scrapbook which wraps around to the beginning if the user tries to move past the last "page", the slide-bar has definite start and end points. These are marked by "slides" which are grey, and which cannot be selected as items (figure 4.20). Any attempt to move the slide-bar further past the end points results in a spoken message from the bird character telling the user to "click the other arrow to go the other way".







fig. 4.20 - three views of the slide bar as the left-hand-arrow is clicked repeatedly (i.e. the slides move to the left). Note in the last view the right-most item is a grey, empty slide which cannot be selected.



There are two ways to implement scrolling, based on different physical models of the relationship between the window and the data. These are referred to as the "moving data", and "moving window" models. The individual slides in the slide-bar scroll in the direction of the arrow that is clicked, an example of the "moving data" model. The window stays put and data "moves" underneath it. This directly conflicts with the standard Macintosh scroll-bar control, such as found on word-processor documents, which scrolls the contents of the window in the opposite direction (figure 4.21, and 4.22). This alternate scrolling form is the "moving window" model. The window "moves" to reveal the data underneath it.


fig. 4.21 The window contents are too large to view all at once and are truncated on the right.



fig. 4.22 Clicking the right-hand arrow causes the window contents to scroll left.



Although anecdotal evidence [Billingsley, 1988] suggests that moving window" is the user's preferred windowing mode, it is likely that this is due to the metaphor of the windowing system. The slide-bar was designed to scroll the slides in the same direction that the arrow points, since it was thought that this was likely to be more intuitive for the target audience. The metaphor for the slide-bar is that of a holder for slides. In order to see slides off to one side the existing slides move out of the way. This suggests that the "moving data" model more accurately reflects the metaphor. This is one theory that remains to be tested, since the tool has not been extensively piloted on children.
back to Chapter 3. forward to Chapter 5