Deliverables Report

IST-2001-33310 VICTEC

<July 2003>

Specification of the bullying Demonstrator

AUTHORS: Ana Paiva, Carsten Zoll, Daniel Sobral, Isabel Machado, Harald Schaub, Li Jin, Lynne Hall, Nisa Silva, Raquel César, Rui Ferreira, Ruth Aylett, Sara Coimbra, Sarah Woods, Sandy Louchart, Sibylle Enz

STATUS:Final

CHECKERS: Sue Richardson, Sandy louchart

PROJECT MANAGER

Name: Ruth Aylett

Address: CVE, Business House, University of Salford, University Road,, Salford, M5 4WT

Phone Number: +44 161 295 2922

Fax Number:+44 161 295 2925

E-mail:

TABLE OF CONTENTS

PURPOSE OF DOCUMENT

EXECUTIVE OVERVIEW

1. INTRODUCTION

2. PRAGMATIC REQUIREMENTS

2.1 Generic requirements

2.2 Pedagogic requirements

Requirements regarding the Bullying Domain

Location-specific requirements

User-specific requirements

2.3 First Experiences / Studies

3. USE CASE OF THE DEMONSTRATOR

Preparation for the Interaction

Interaction with the Application

End of interaction with Demonstrator: Questionnaires

4. IMPLEMENTATIONAL APPROACH WITHIN VICTEC WORK

4.1 Toolkit Integration

4.2 Framework

4.1.1 Characters

4.1.2 Language

4.1.3 Narrative

4.1.4 Interaction and Sentoy

4.7 Evaluation

5. CONCLUSIONS AND FURTHER WORK

5.1 Work done so far

5.2 Further Work

6. REFERENCES

ANNEX 1 - A PSYCHOLOGICAL OVERVIEW OF A SESSION WITH THE DEMONSTRATOR

ANNEX 2 – CHARACTERISTICS FOR EACH BULLYING PROFILE

ANNEX 3 – BULLYING SCENARIO SPECIFICATIONS FOR DIRECT, VERBAL AND RELATIONAL BULLYING BEHAVIOUR

ANNEX 4 – CATEGORIES OF COPING RESPONSES TO DEAL WITH BULLYING SITUATIONS AND CORRESPONDING SUCCESS RATES

ANNEX 5 – DESIGN PROTOCOL FOR QUESTIONNAIRES

ANNEX 6 – WIZARD OF OZ STUDY IN A PORTUGUESE SCHOOL

ANNEX 7 – PROTOTYPICAL SCHEMATIZATION OF THE DEMONSTRATOR SOFTWARE ARCHITECTURE

PURPOSE OF DOCUMENT

This document constitutes Deliverable 6.3.1 of the IST Project VICTEC, on the Specification of the Bullying Demonstrator. It relates the requirements that psychologists and experts on education have for the intended functionality of the final version of the Demonstrator. Further, it proposes a technical approach that tries to provide solutions to achieve the described requirements. Finally, since the demonstrator is the final product of this project, it also shows the relationship between the other components of the project and the Demonstrator.

EXECUTIVE OVERVIEW

This Deliverable is the first deliverable from WP6 (Specification of the bullying demonstrator) of the IST-sponsored VICTEC project. This Workpackage has as main goal to demonstrate the use of the technology developed in the other workpackages through the construction of a virtual improvisational drama environment dealing with violence and bullying in schools. This document specifies in greater detail the goals this workpackage is trying to achieve and proposes some technical solutions to achieve them, regarding the work done in the other workpackages.

1. INTRODUCTION

The anti-bullying demonstrator is one of the main products of this project, and therefore relates to all the project aims, as described in the Technical Annex, part 1, p. 2. Particularly, it directly concerns the following points:

P1: Allow children to explore perspective taking and reflection for dealing with complex social problems (Task 3, WP5; WP6).

P4: To demonstrate to the education and training community technology for educational applications in personal and social education using a new kind of technology (VEs) and in particular virtual improvisational drama. (WP6)

This demonstrator will have a multifaceted role in this project:

• To prove experimentally the validity of the research conducted on the development of synthetic characters and the “empathic relation established” between children and such characters.

• To illustrate the use of the framework in innovative and practical examples, where empathy, believability and credibility of synthetic characters play a fundamental role.

The demonstrator takes the form of an episodic virtual drama for which the content is set up using the Toolkit discussed in D3.2.1 and delivered in prototype form in D3.2.2. A single child user interacts with the demonstrator by acting as the ‘invisible friend’ of one of the virtual characters and advising them between episodes. An introduction segment is located at the start of the episodes, and a reflective interactive segment at the end. See section 3 for a diagrammatic representation of this structure.

Section 2 discusses the psychological and educational requirements. Section 3, Use Case of the Demonstrator, provides a brief example of the expected interaction with the Demonstrator, as envisioned at this stage. In section 4, Implementational Orientation within VICTEC work, we propose a technical solution that reflects all the requirements described. As the Demonstrator is a global result of the entire project, we also discuss the integration of all the project components.

It should be noted that the project is applying an iterative prototyping methodology to its software outputs as the most appropriate approach in a domain where most of the issues are the subject of research. This specification cannot therefore be taken as the ‘last word’ in the way commonly assumed where a waterfall approach is taken. It is likely that some modification of the specification will occur as the result of experience in developing the demonstrator and early evaluations of it.

-1-

Deliverable 3.1.1/Report no.1/Version 1

Deliverables Report

IST-2001-33310 VICTEC

<July 2003>

2. PRAGMATIC REQUIREMENTS

Some of the issues regarding the evaluation of software being produced in the project (Demonstrator and Toolkit) have been discussed in Deliverable D2.2.2, Evaluation Methodology. A brief overview of the requirements for the Demonstrator will be discussed in this section. Although it is not mandatory, a previous reading of D2.2.2 can help in clearing some of the ideas present in this section.

2.1 Generic requirements

The Demonstrator must create a safe virtual environment depicting school situations. A semi-realistic (see Deliverable D5.1.1 for further discussion about realism) 3D environment is essential to achieve this. Active participation of the user is required to support the educational process and produce its results. Therefore, it must support an interactive capacity, and avoid the mostly scripted nature of the majority of educational systems.

The Demonstrator must be generally usable within the school context. Therefore, it must run on the simple PC machines which most schools have, often with limited processing capacity and without the ability to support any specific hardware. Albeit the fact that 3D applications are very demanding on a computer, current systems available in schools have graphics cards that we believe will be adequate. Nevertheless, these limitations must be taken into consideration in the design of the software.

Interaction other than via mouse and keyboard is likely to be infeasible on the large scale. The limited number of Sentoys (see Deliverable 4.1.1 for further details) and difficulties in the generalization of other interactive methods will allow only limited (and laboratory-based) studies enabling further interaction capabilities.

2.2 Pedagogic requirements

Most of the requirements relating to the strong psychological and pedagogical components of the project have been thoroughly analysed and discussed in other Deliverables, namely D2.1.1 (Learner Scenarios: Scenarios Requirement Definition). However, due to their inherent importance, we will summarise them again here.

Requirements regarding the Bullying Domain

A key psychological and educational goal for the demonstrator is to provide child users with the opportunity to work through a variety of different bullying situations that occur in real-life settings for both direct/physical and relational bullying. Relational bullying scenarios pose a greater challenge for the demonstrator due to their high language content and the discrete use of body language (e.g. shunning people, turning your back on people, ‘dirty’ looks). Due to the sensitivity of the topic, multiple users of the same scenario are contra-indicated, because of the psychological stress that might be caused to a victimised child. Indeed an advantage of the demonstrator over other components of the teaching repertoire for this topic is seen as the ability for a child to explore the issue individually.

Annex 2 illustrates the profiles for the main characters that will be represented within the demonstrator: ‘pure’ bullies, ‘pure’ victim, bully/victims and neutral children. Defenders (children who help the victim) and Bystanders (children who passively watch bullying events) are classified as neutral children and need to be represented as character profiles within the demonstrator. These specifications will assist the technical team with building personality profiles, action and language repertoires for each character. Deliverable D5.1.1 ‘Specification of Empathic Synthetic Characters’ discusses the architecture that will support these profiles.

Location-specific requirements

The Demonstrator must be implemented in three different languages (English, German, Portuguese). This will impose an extra burden in the language component of the software (for more details see Deliverable D3.2.1 and Section 4.1.2 Language).

Further serious issues also arise from the different cultures involved. We expect to overcome this through the authoring system (see D3.2.1 and D3.3.1) which allows the definition of location-specific scenarios (see D2.1.1 for a further insight on culture-specific scenarios). Nevertheless, these location-specific scenarios will demand location-specific resources (eg., different schools and props for rural and urban sets, uniforms in England) . This, unlike the language issue, will impose further creative design effort but will hopefully have little implications on the software design of the Demonstrator.

User-specific requirements

It is essential that the demonstrator has specifications which allow different bullying stories and episodes to be selected for each Virtual Learning (VL) session depending on the child user characteristics. From a psychological and educational perspective, we do not want the demonstrator to make competent bullies more skilled by providing new perspectives and ideas. This is a key challenge for the VICTEC project and it is important that the scenarios designed for the demonstrator are high quality so that they produce empathic feelings between the child user and the victim characters in the scenarios. We do not want to portray bullies as having a ‘cool’ image within the scenarios and it is important to avoid heavy reliance on character stereotypes for bullies, bully/victims, victims, neutral children (bystanders or defenders). The authoring tool is also essential in achieving this objective.

An important psychological and educational requirement of the demonstrator is to allow children to explore and try out different types of coping responses to deal with physical and relational bullying situations. From a psychological viewpoint, we are also interested in examining the rationale and justifications that children provide for different coping responses (e.g. do children who are bullies select different coping strategies and rationale than users who are victims?). There are a diverse number of different coping responses available to deal with bullying behaviour that have different probabilities of being successful or unsuccessful dependent on factors such as age, gender, physical appearance (weak or strong, tall or small), personality, number of friends, type of bullying behaviour, and past experiences of bullying behaviour. In Annex 4 we summarise the range of coping strategies available to children and their likely success/failure rates.

The mechanism by which children select a coping mechanism has been extensively discussed. A simple solution would be to select from a menu of coping responses. However there are strong arguments against this: it is an unnatural way of interacting with a character, but more than this, the Wizard of Oz test discussed in Annex 6 suggested that the child would nearly always choose the ‘socially acceptable’ answer of Tell Someone. The demonstrator would then fail to meet its objective of allowing exploration of strategies.

An alternative would be to give the child complete freedom to choose and write something. This is technically more difficult to implement and may put a heavy reliance on the child’s typing skills. The current feeling is that children should be able to freely select what coping response they would like, but that if they keep selecting inappropriate or the same response, they could be told that they can visit a ‘resource room’ where further coping responses will be available to select. This relates to the technical issue that deals with language (Section 4.1.2 Language).

Gender is an important consideration for the demonstrator in terms of matching the gender of the user and the characters in the scenarios. This is particularly salient for boys. Preliminary results from a study examining gender preferences with animated characters depicting a physical and relational bullying scenario revealed that boys significantly preferred the male victim character, whilst girls were less gender specific (58% preferred the female victim & 31% preferred the male victim). A gender trend was also found for least liked characters in bullying scenarios: boys least liked the female bully characters whilst girls were more evenly spread (48.5% least prefer the male bully & 42% least prefer the female bully). When children were asked which character they would like to be, boys did not want to be girl characters, particularly girl bullies. A significant association was also found between gender and same gender victim: 86.2% of boys wanted to be the male victim and 81.5% girls wanted to be the female victim. A gender effect was also uncovered for empathic feelings expressed towards the characters in the bullying scenarios. Boys expressed affective empathy for the male victim but not the female victim, whereas girls expressed empathy for both the male and female victim. These findings clearly illustrate the need for the demonstrator to consider the gender of the user and the gender of characters in the scenarios to ensure full engagement and encourage empathic relations between the user and the characters. (Woods, Hall, Sobral, Dautenhahn & Wolke, 2003).

2.3 Initial Experiences / Studies

As stated above, the project follows a process of iterative steps of increasing functionality. This is especially needed where, as in the case of VICTEC, most of the underlying concepts are still under investigation.

In order to clarify aspects of the scenarios, characters and overall functionality desired for the final application, some initial experiences and studies have been conducted. Some of the conclusions regarding the characters and scenarios are analysed in deliverables D4.1.1 (scenario requirements) and D5.1.1 (character specification). Further studies have already explored a 3D environment (Woods et al., 2003), carrying out a ‘proof of concept’ exercise of the technology for the purposes it is trying to achieve.

As already mentioned, Wizard of Oz studies have also been conducted (see Annex 6 for an example conducted in a Portuguese school). In a Wizard of Oz study, the user is made to believe that they are interacting with a system, while in reality they are interacting with a human Wizard, sitting behind the “screen” pretending to be the system. This method can be useful in situations where the development of the system is expensive and where it is hard to know beforehand how users will behave as there are no “natural” behaviour experiences or given user tasks to build upon. The method has been used in the design of natural language systems (Dahlbäck 1993) and intelligent agents (Maulsby 1993)One of the main advantages is that the method does not constrain users’ possible expressions.

The main focus of the study described in Annex 6 is the interaction between the user and the characters, namely how the user can advise the victim in defining possible coping responses.

We expect these studies to provide insights into users’ response patterns. Furthermore, they also contribute to usability studies relating to the ability of the users to convey their thoughts and feelings in an appropriate manner. Such studies will continue as part of the iterative prototyping approach being taken.

-51-

Deliverable 3.1.1/Report no.1/Version 1

Deliverables Report

IST-2001-33310 VICTEC

<July 2003>

3. USE CASE OF THE DEMONSTRATOR

Each scenario within the demonstrator be expressed through up to 5 episodes during which the child user will have the opportunity to try out a number of different coping responses to deal with different bullying situations that arise. Due to the non-scripted nature of the demonstrator and the use of emergent narrative, no one scenario will be the same and different episodes will be shown with some randomness. See Deliverable 5.1.1 (Specification of Empathic Synthetic Characters) for more detail on the narrative mechanisms. In Annex 1 we can see an illustration of a psychological overview of the requirements for a session with the demonstrator.

In the following subsections a brief overview of a session with the Demonstrator is described. Figure 1 shows a brief flow diagram of this session.

Figure 1: Flow diagram of a session with the Demonstrator

The session will start with an off-line set of Questions and Answers preparing for the interaction, which starts with a message displaying the context under which the scenario will happen. Then, the scenario will develop, with a sequence of five episodes, which will consist of a series of encounters between the characters in specific situations. Between each episode, where bullying situations are expected to happen, a coping dialogue between the user and the victim occurs, where the user should advise the victim about possible solutions to the bullying situation he/she just witnessed. After the five episodes, a final message is displayed, along with a visual storyboard of the episodes displayed. Finally, another off-line questionnaire phase will occur performing the necessary psychological/pedagogical assessment.

Preparation for the Interaction

A successful demonstrator must show different types of bullying scenarios within the right school environments and with closely matched character profiles. There are three main types of probable bullying scenarios, 1) direct/physical, 2) verbal, 3) relational. Different types of bullying behaviour are more prevalent in certain environments compared to others and are likely to have different character profiles. Annex 3 illustrates bullying scenario specifications for direct/physical, verbal and relational bullying behaviour. As we have mentioned, due to the inherent complexities for the language system, we are currently focusing on the direct/physical scenarios, although we must address all different scenarios.