Instructor Support in Collaborative Multiplayer Serious Games for

Learning

Game Mastering in the Serious Game ’Woodment’

Viktor Wendel, Michael Gutjahr, Stefan G

obel and Ralf Steinmetz

Multimedia Communication Labs - KOM, TU Darmstadt, Rundeturmstr. 10, 64283 Darmstadt, Germany

Keywords:

Serious Games, Collaborative Learning, CSCL, Game Mastering, Adaptation.

Abstract:

In collaborative digital learning scenarios with small groups (3-6 users), the role of the instructor is vital as

he/she is responsible for preparation of the setting, observation, coaching, moderation and adaptation. Cur-

rently, in multiplayer Serious Games, the role of the instructor is only insufﬁciently considered. Only very few

approaches for integrating or supporting instructors in collaborative multiplayer Serious Games exist today,

to the best of our knowledge. In this paper, we propose a concept for integration and support of instruc-

tors in team-based collaborative 3D multiplayer Serious Games. Our approach is based on Game Mastering

principles known from roleplay games. It combines those principles with concepts for collaborative learning

scenarios. We applied our concept to the existing 3D multiplayer Serious Game Woodment and tested it in a

vocational school with 26 players in four groups (age: m= 19.12; sd= 2.03). Results indicate that an instruc-

tor using our Game Master framework to moderate and adapt the game at runtime can have a positive effect

on both the players’ learning success and perceived user experience. Moreover, a positive effect on players’

gaming behavior can be observed.

1 MOTIVATION

Although many promising examples of Serious

Games for learning are existing today, most of those

are single player games.This is true even for social Se-

rious Games (Konert et al., 2012) which are usually

played alone despite asynchronous interaction with

other players/friends. We believe that multiplayer Se-

rious games can be a promising opportunity to com-

bine the advantages of Serious Games with the con-

cepts of collaborative learning, especially Computer-

supported Collaborative Learning (CSCL). However,

in (computer-supported) collaborative learning sce-

narios, the role of the instructor is vital. (H

ainen

et al., 2006) argue that ”Computer-supported Collab-

orative Learning (CSCL) must provide instructional

support”. The instructor is responsible for various

tasks in and around the learning process. According

to (Haake et al., 2004), page 30, those are, among

others, analysis, monitoring, moderation, guidance,

coaching, and intervention. However, to the best of

our knowledge there are only very few approaches to

support an instructor in digital game-based collabora-

tive learning scenarios.

Our approach uses concepts from Game Mas-

tering (in pen&paper roleplay games), collaborative

learning, and collaborative gaming. In (Wendel et al.,

2012a), as a ﬁrst step, we proposed a model-based ap-

proach for a Game Master concept in 3D Multiplayer

Serious Games. The paper focused on the interface

between the game and the Game Master frontend as

well as on a group model consisting of a player model,

a learner model, and an interaction model. The main

contribution of this paper is an extension of our model

by proposing methods and concepts for providing a

Game Master with the ability to analyze, monitor,

coach, intervene, and adapt at runtime from inside the

game. Therefore, we will describe methods for get-

ting required information from the game to perform

those tasks as well as for providing adequate methods

of adaptation of the game. We implemented our ex-

tended concept as an extension to the existing collab-

orative multiplayer Serious Game Woodment (Wen-

del et al., 2010) and performed a user-centered eval-

uation at a vocational school in Germany. Our hy-

pothesis is that an instructor using our approach can

positively inﬂuence both game experience and learn-

ing outcome of players in an immersive way. The

study was conducted at a vocational school in Ger-

many with 26 students (age: m= 19.12; sd= 2.03).

Wendel V., Gutjahr M., Göbel S. and Steinmetz R..

Instructor Support in Collaborative Multiplayer Serious Games for Learning - Game Mastering in the Serious Game ’Woodment’.

DOI: 10.5220/0004836900490056

In Proceedings of the 6th International Conference on Computer Supported Education (CSEDU-2014), pages 49-56

ISBN: 978-989-758-022-2

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

Results show that an instructor using our Game Mas-

ter frontend to moderate and adapt the gaming process

at runtime can have a positive effect on both the learn-

ing success and the perceived user experience of the

players. Moreover, different effects on the gaming be-

havior of players can be observed like a more focused

play style and a more coordinated teamwork.

2 RELATED WORK

The concept of Computer-supported Collaborative

learning (CSCL) is being used in various learning

scenarios ranging from learning at school, university

learning, to training scenarios in corporate environ-

ment (Shell et al., 2005), (Stahl et al., 2006), (On-

rubia and Engel, 2009), and (Larusson and Alterman,

2009). As stated by (Kearsley, 2000) or (Chiriac and

Granstr

om, 2012), the instructor’s role is vital in col-

laborative learning scenarios. Instructors have vari-

ous important tasks during collaborative learning ses-

sions. (Mutwarasibo, 2013) states that “the instruc-

tor’s role in student group work is that of a guide or

facilitator”. A more detailed description of instructor

tasks in collaborative learning scenarios, especially in

CSCL scenarios is given in (Haake et al., 2004), page

30. There, the tasks of an instructor are described as:

analysis, monitoring, moderation, guidance, coach-

ing, and intervention.

Grand challenges of multiplayer Serious Games

research are: multiplayer Serious Game design, in-

teraction and communication on game-based collab-

orative learning scenarios, and the role of the in-

structor as well as instructor support. (H

ainen

et al., 2006) describe a concept for designing collab-

oration in a 3D virtual game environment. The role

of the instructor concerning real-time orchestration in

a 3D game is discussed in (H

ainen and Oksa-

nen, 2012). Design guidelines for incorporation of

features of collaborative learning in video games are

presented by (Zea et al., 2009). (Wendel et al., 2010)

describe the design of a collaborative multiplayer Se-

rious Game for collaborative learning. The work pre-

sented here is based on this game.

The concept of Game Mastering as it will be used

in this paper, is derived from pen&paper roleplay

games. In such games, the Game Master (GM) is

responsible for telling a suspenseful and interesting

story to a group of players which are active parts in

that story. Thus, they are able to actively inﬂuence

that story, oftentimes against the GM’s plan. The

GM’s task is to keep the story meaningful while at

the same time allowing a maximum of freedom for

the players without giving them the feeling that their

actions are futile. Thus, it is the GM’s task to unite

those goals (well considered story vs. player free-

dom). This problem is often referred to as the Nar-

rative Paradox (Louchart and Aylett, 2003). There-

fore, the GM needs to be able to observe the game in

the context of the whole story and be able to adapt

his/her plan ad-hoc according to the needs and prefer-

ences of the players. These requirements are similar

to those of an instructor in a (game-based) collabora-

tive learning scenario. First concepts to formalize the

work of a Game Master in order to transfer the Game

Master concept to digital games have been proposed

by (Tychsen et al., 2005) and (Tychsen, 2008).

Various concepts for modeling player behavior

and preferences have been proposed up to today. The

most prominent player model is Bartle’s model for

roleplay types (Bartle, 1996). A more generic model

has been proposed by (Houlette, 2004). A compre-

hensive overview over player models is provided in

(Smith et al., 2011). They provide a taxonomy of

player modeling. In terms of learner modeling, the

concept of (Kolb and Kolb, 2005) is one of the most

prominent concepts. However, for the scope of this

work, we focus on modeling the learning progress

of learners. Therefore, we need to model what a

player/learner has already learned. It is also desirable

to be able to make suggestions for the next learning

unit based on already learned content. Therefore, a hi-

erarchical approach to modeling learner skills seems

suitable like proposed by (Korossy, 1999).

3 OUR APPROACH

The approach presented in this paper is based on our

concept for Game Mastering as described in (Wen-

del et al., 2012a). There we propose a component-

based model for describing a 3D multiplayer Serious

Game. Our concept deﬁnes a ’group model’ to model

players/learners. The group model is composed of

a player model, a learner model, and an interaction

model. In this paper, we want to extend that model

about a client sided interface declaring how informa-

tion from a game can be extracted to be presented to

an instructor in a useful manner. Furthermore, the ex-

tended model deﬁnes an interface for accessing rel-

evant game entities in order to be able to adapt the

game at runtime.

3.1 Architecture

The architecture of our framework (see Figure 1) con-

tains two main elements: One is the 3D game itself,

the other one is the Game Master frontend. Via the

CSEDU2014-6thInternationalConferenceonComputerSupportedEducation

Figure 1: GM Framework Architecture

deﬁned interface (between game and frontend), the

Game Master frontend can access relevant data from

the game and on the other hand adapt game rules or

entities. The Game Master can use the GM frontend

to receive necessary information and to inﬂuence the

game according to his/her professional knowledge in

order to optimize learning success, gaming experi-

ence, and collaboration. The interface is split into four

sub-components: Entities Compound, Group Model,

Information Model, and Adaptation Model. The com-

ponents will be explained in more detail below.

3.2 Entities Compound

The Entities Compound deﬁnes all relevant game en-

tities. According to (Wendel et al., 2012a), 3rd person

3D games consist of three main parts: Game World,

Players, and Interaction. The Entities Compound de-

scribes the former two of those. It deﬁnes all relevant

game entities. Relevant entities are all those game en-

tities which’s state inﬂuences the state of the game.

Those are of course the player entities. Additionally,

all 3D assets which have a purpose beyond setting the

scene, i.e. background images, static 3D objects, and

terrain. Generally, those are all game entities a player

can interact with in any way. It is task of the game de-

signer to deﬁne which game entities are relevant and

which ones have only a decoration purpose. A game

entity exists of a description, describing the function

of the entity in the game, and a list of parameters. For

each parameter the game designer needs to deﬁne if

it is only informational or if it can be changed. Addi-

tionally, a parameter description needs to be provided,

explaining the parameter’s function for the entity.

3.3 Group Model

The Group Model contains the state of the group of

learners. It is described in detail in (Wendel et al.,

2012a). It provides information about player behav-

ior and preferences in play style (player model). Fur-

thermore, it describes the learning progress for each

player. Finally, it contains information about interac-

tion and communication between players.

The Player Model describes preferences in gam-

ing behavior for a player, e.g. if a player prefers ac-

tion, socializing with other players, or wants to ex-

perience every aspect of a game, i.e. ﬁnd every hid-

den piece of it. The most common and best ﬁt player

model for those RPG-similar 3D multiplayer games is

the player model of Bartle (Bartle, 1996). It classiﬁes

players along two axes (players - world, acting - inter-

acting) resulting in the four stereotypical player types:

killer (player, acting), achiever (world, acting), social-

izer (player, interacting), and explorer (world, inter-

acting). Again, the game designer(s) need to specify

which action and decision a player can take during

the game relates to which player type. In our model,

this is done by assigning player model modiﬁcation

values to relevant player actions.

The purpose of the Learner Model is to give the in-

structor a structured insight into the learning progress

of players. Following the hierarchical model of Ko-

rossy (Korossy, 1999), the game designers/subject

matter experts deﬁne a set of skill which will should

be learned throughout the game. Those are ordered

in a hierarchical structure indicating dependencies be-

tween skills. A skill depends on another skill if that

other skill should be learned before this skill is looked

at.

The Interaction Model is meant to provide the in-

structor with information about interaction between

InstructorSupportinCollaborativeMultiplayerSeriousGamesforLearning-GameMasteringintheSeriousGame

'Woodment'

players. This contains any means of communication.

Usually, this includes a chat protocol. A more de-

tailed look into communication between players con-

tains statistics about the frequency of communica-

tions between players. Apart from communication,

interactions are taken into account. Therefore, the

game designer needs to deﬁne which actions in a

game taken by a player are an interaction with an-

other player. This is similar to the actions deﬁned for

the player model.

3.4 Information Module

Apart from collected data like the group model, it is

useful for the instructor to be able to directly observe

the gaming progress (Haake et al., 2004) in order to

be able to extract information about the collaborative

learning process. Moreover, it might be useful for

recognition of problems players might have at certain

points in the game. Therefore, it seems useful for the

GM to be able to move freely in the game world, i.e.

have a free camera perspective. Moreover, in order

to be able to spectate what all players are doing at a

certain point of time when players are split up, it is

useful to provide a split-screen camera. Apart from

this, the GM should be provided with the information

of the group model. Finally, general game parameters

as well as information about the state of game entities

should be displayed.

3.5 Adaptation Module

The Adaptation Module is the part where the instruc-

tor can inﬂuence the game by adapting game param-

eters, adapting parameters of an entity, and adapting

a game rule.

Therefore, it needs access to the basic game pa-

rameters as well as to the game entities. Thus, it is

connected to the Entities Compound. In addition to

that, it needs to access the rule base of the game. As

game rules signiﬁcantly inﬂuence a game on a very

basic layer, the game designer needs to deﬁne which

rules may be changed in which way. For example, it

could be allowed to perform a certain action or not. In

order to simplify this access, an abstraction interface

will be put between the game rule base and the Adap-

tation Module. Changing game rules can be imple-

mented through adapting a set of (boolean) parame-

ters, provided game rules are designed carefully. Note

that learning parameters, like difﬁculty of questions,

etc., are capsuled in game parameters or parameters

of (learning) entities. However, they should be dis-

played to the GM in a suitable way separating them

from gaming parameters. Thus, via the adaptation

module, the GM is able to manipulate relevant 3D

objects, game rules (i.e. interaction rules, rules for

collaboration, game actions), and difﬁculty in terms

of gaming, or learning.

4 PROOF-OF-CONCEPT

4.1 Woodment

As a proof-of-concept, we implemented our approach

as an extension to the existing Serious Game pro-

totype Woodment. The game has been chosen be-

cause it has a rather high level of interaction and a

trainer/teacher can freely deﬁne the learning content.

It is possible to create, save, and reuse question sets

of various content domains and even to mix them if

so wished. Woodment is a 3D 3rd person multiplayer

digital educational game for 6 players. The game

has been developed by the authors and been enhanced

during practical courses or theses. One enhancement

of Woodment was performed as a part of a diploma

thesis which focused on Game Mastering concepts

and team leadership (Rodenberger, 2012). Some of

the visualization concepts presented in this paper have

been reimplemented from this work.

Woodment as well as the enhancements imple-

mented as a proof-of-concept for the work presented

here, was implemented using the Unity3d game en-

gine. The game can be played both in browser (Unity

browser plugin required) and as a standalone applica-

tion (for PC and Mac), providing us with the neces-

sary platform independence for evaluations on vary-

ing hardware in different vocational schools.

4.2 Game Entities

Question orbs contain questions which can be trig-

gered when a player enters the orb. They provide a

game relevant resource (food, workers, or ship tokens

to sell wood) if the answer is correct. The Game Mas-

ter can adjust their size and movement speed as well

as the frequency each question orb type spawns with.

Skill canisters can be picked up in order to be able to

run faster, freeze an enemy player for a period of time,

or to ignite the enemy base forcing the enemy team to

spend time to extinguish the ﬁre. The GM can adjust

their spawning frequency. Workers gather resources

provided that they have sufﬁcient food. They cannot

be accessed directly by the GM. The player base is

the center of operations for a team. The GM can view

it and see if it is currently burning, but not directly

adjust that fact.

CSEDU2014-6thInternationalConferenceonComputerSupportedEducation

4.3 Group Model

We deﬁned several player actions like ’Pickup Skill

Canister’, ’Trigger Question’, ’Answer Question Cor-

rectly’, ’Ask For Help’, or ’Help Player With Ques-

tion’, which affect one or more of the player model

items (killer, achiever, socializer, explorer).

Woodment contains an integrated editor enabling

the instructor to create a custom set of questions.

The set of questions as well as dependencies between

questions are created by the instructor according to

the learning content. Thus, it is possible to model the

learning space according to the Competence-based

Knowledge Space theory (Korossy, 1999). The ac-

tual learner modeling is done by modeling the learn-

ers’ initial knowledge and monitoring the acquisition

of knowledge during the gaming phase and updating

the learners’ knowledge state accordingly.

All player chat is directly visible to the GM. In ad-

dition, player chats are aggregated such that for each

player it is known how often he/she communicated

with each other player. This is presented to the GM

graphically. In terms of interaction, the game records

whenever a player asks for help or gives help for a

question. Also, whenever a player freezes an op-

ponent or de-freezes a fellow player, the interaction

counter for the affected players increases. This is also

presented graphically.

4.4 GM Observation Frontend

To satisfy the needs for observation of the gaming oc-

currence, we implemented a ’ﬂy mode’ for the GM,

i.e. the GM can move freely within the game world.

This includes a smooth zoom. Moreover, the GM

can automatically follow each player by clicking the

player’s name. A split screen mode was not imple-

mented as dividing the screen into 6 parts seemed

to be too confusing. The group model is displayed

in a special window providing overview over each

player’s player model, learner model, and interaction

model. The GM can view game entities directly in the

game world and see their parameters in the settings

window.

4.5 GM Adaptation Frontend

The GM can adapt game entity parameters (if al-

lowed) in the settings window. Moreover, the GM

can directly adjust the most signiﬁcant game parame-

ters directly on the main screen. Those are the num-

ber of wood gathered, workers, ship tokens, food, or

gold for each team. Further, the GM can adapt some

visual options like sunlight, brightness or fog which

indirectly inﬂuence the game (difﬁculty). Poor sight

makes ﬁnding question orbs harder which increases

the need for communication among team members.

5 EVALUATION

5.1 Hypothesis and Setup

Our Hypothesis is that an instructor is able to sup-

port and positively inﬂuence the collaborative learn-

ing process as well as the game experience of a gam-

ing session, especially the game ﬂow. A Game Master

(GM) frontend is implemented, following the concept

presented in this paper. Therefore it is hypothesized:

1. Comparing with a scenario without a GM, a GM

increases the ﬂow experience and user experience.

2. Comparing with a scenario without a GM, a GM

increases the learning success.

Participants were playing one of two versions of

Woodment. The treatment group was playing Wood-

ment with support of a GM, while the control group

was playing the standard version without the support

of a GM. Both groups were playing Woodment for

40 minutes and answered a questionnaire after play-

ing the game. The questionnaire was a modiﬁed ver-

sion of the user experience questionnaire described

in (Wendel et al., 2012b) and includes an overall user

experience (UX) score, as well as seven UX sub-

scales. The in-game answers to the content of the

curriculum were logged, too, as well as all player and

GM actions. For both groups (treatment and control

group), two gaming sessions with 6(7)

players each

were conducted. Players were from different classes

of a vocational school in Germany. The study in-

cludes 26 participants, 18 male, 5 female and 3 not

stated. Age m= 19.12 (sd= 2.03). To analyze the data

an ANOVA between subjects was used. The present

of the GM was used as independent variable and the

overall scale, as well as the seven sub-scales were

used as independent measurements.

5.2 Results

The GM version (m=6.14; sd=1.44) triggered more

overall user experience (F(1,23)=6.93; p=.015) than

the control group without a GM (m=4.90; sd=0.89).

To detect which aspects of user experience are es-

pecially distinct in the GM version, the seven sub-

scales were tested, too. The sub-scales Immer-

sion (F(1,23)=7.54; p=.012), Flow (F(1,23)= 11.50;

due to class sizes, in two groups we had 7 players, two

students each played together on one computer

InstructorSupportinCollaborativeMultiplayerSeriousGamesforLearning-GameMasteringintheSeriousGame

'Woodment'

Figure 2: Questions answered by player

p=.003), and Arousal (F(1,23)= 4.25; p=.051) showed

an effect in favor of the GM version.

During the gaming sessions, the game logged all

relevant data. Logged data contains all GM and player

actions, triggered questions and answers, chat data

and all game variables logged in a one second in-

terval. From the post-processed data we could see

the progress of resources (wheat, ﬁsh, workers, idle

workers, lumber, gold) for each team over time

Moreover, we aggregated the data for the questions

answered (see Figure 2) by the players and the player

models (see Figure 3).

Figure 2 shows the number of questions triggered,

and answered correctly or wrong for each player.

More questions (F(1,22)=6.77; p=.016) have been

solved correctly in versions with a GM (m=13.25;

sd=4.81) than in versions without a GM (m=7.83;

sd=5.37). Also, the percentage of correctly solved

questions is higher (F(1,22)=7.53; p=.012) in ver-

sions with a GM (m=61.21; sd=14.24) than in ver-

sions without a GM (m=41.58; sd=20.29). As an in-

dicator for the overall success of a team, we looked

at the gold they were able to achieve. Without a GM,

the teams got 2/2 (red/blue) and 2/3 (red/blue) gold

(9 gold total among all four teams). The teams with

For reasons of clarity and available space, we cannot

include all of those plots in this paper. However, they are

available from the authors upon request.

a GM got 3/3 (red/blue) and 2/4 (red/blue) gold (12

gold total).

5.3 Discussion

Results show that both UX questions and the percent-

age of correctly answered questions are signiﬁcantly

larger in the game sessions where the GM was present

and used the GM frontend. Thus, we can accept both

hypotheses.

Comparing the player models, it comes to the

fore in the sessions where a GM was present, play-

ers were generally more active. The values for ex-

plorer (232/254 with GM, 158/152 without GM) show

that players tend to move more when a GM is present.

This indicates that players were more actively search-

ing for question orbs or supporting fellow players

(e.g. de-freezing). However, as we cannot track the

purpose of a movement, it needs to be clariﬁed that

higher explorer values can also mean that the respec-

tive player might just be moving around for other rea-

sons. Achiever values (350/333 with GM, 264/238

without GM) indicate that players play more com-

petitive when a GM is present. Collaboration values

do not differ greatly (84/77 with GM, 50/84 without

GM). Killers are also more active (100/134 with GM,

73/93 without GM) when the GM was present. Over-

all the activity (total number of actions performed)

CSEDU2014-6thInternationalConferenceonComputerSupportedEducation

Figure 3: Player Model

was higher in those games where the GM was present

(766/798 with GM, 545/567 without GM).

From the observations described above, we con-

clude that the GMs were able to make students focus

on their tasks by reducing disturbing behavior and by

recognizing problems in teams. Moreover, it seems

that GMs were able to recognize problems/errors at

answering questions and that they were able to pro-

vide useful help, thus increasing both the average

number of questions triggered, and the amount of cor-

rectly answered questions. This indicates that the GM

was able to fulﬁll his/her traditional tasks like obser-

vation or coaching, and help players learn better while

improving UX.

5.4 Limitations

The ﬁndings of this paper are limited by the follow-

ing constraints. The game sessions have been con-

ducted with only a small number (26) of participants

of which 18 ware male. Although the students en-

thusiasm towards playing the game and their motiva-

tion during the gaming time was seen as very positive

by their teachers, it needs to be considered that this

might at least partially result from the fact that play-

ing a game was just a more fun alternative to regular

class sessions. Our concept was only evaluated using

one implementation for one game (Woodment), thus

a general validity for all types of collaborative multi-

player Serious Games can not be derived.

6 CONCLUSION

In this paper we proposed a model for assisting an

instructor in orchestrating a collaborative multiplayer

Serious Game. Our concept is an extension to our

prior work on a framework for Game Mastering in

collaborative multiplayer Serious Games. It deﬁnes

an interface for game developers and subject matter

experts, which can use it to deﬁne relevant game enti-

ties, game parameters, or learning content. We further

provide a concept for a group model based on player

modeling, learner modeling and interaction model-

ing. Our concept addresses a suitable way of present-

ing relevant information to the instructor (GM) and

meaningful ways of adapting the game state and game

rules via the interface. We implemented our concept

as an extension of the collaborative multiplayer Seri-

ous Game Woodment and conducted a user-centered

evaluation with 26 users. Our hypothesis was that an

instructor is able to use a frontend implemented ac-

cording to our model in order to perform vital instruc-

tor tasks in collaborative learning scenarios and sub-

InstructorSupportinCollaborativeMultiplayerSeriousGamesforLearning-GameMasteringintheSeriousGame

'Woodment'

sequently be able to improve player performance in

terms of game experience and learning success. Our

results indicate that our hypothesis can be accepted

with some limitations. Both UX and learning perfor-

mance are signiﬁcantly higher in the gaming sessions

where a Game Master was using the GM frontend

implemented according to our concept compared to

those sessions where no GM was intervening.

REFERENCES

Bartle, R. (1996). Hearts, clubs, diamonds, spades: Play-

ers who suit MUDs. Journal of Virtual Environments,

1(1):19.

Chiriac, E. H. and Granstr

om, K. (2012). Teachers Leader-

ship and Students Experience of Group Work. Teach-

ers and Teaching, 18(3):345–363.

Haake, J., Schwabe, G., and Wessner, M. (2004). CSCL-

Kompendium: Lehr-und Handbuch zum computerun-

terst

utzten kooperativen Lernen. Oldenbourg Wis-

senschaftsverlag.

ainen, R., Manninen, T., J

arvel

a, S., and H

akkinen, P.

(2006). Learning to Collaborate: Designing Collabo-

ration in a 3-D Game Environment. The Internet and

Higher Education, 9(1):47 – 61.

ainen, R. and Oksanen, K. (2012). Challenge of sup-

porting vocational learning: Empowering collabora-

tion in a scripted 3d game - how does teachers’ re-

altime orchestration make a difference? Comp. and

Educ., 59:281–293.

Houlette, R. (2004). Player modelling for adaptive games.

AI Game Programming Wisdom II, pages 557–566.

Kearsley, G. (2000). Online Education: Learning and

Teaching in Cyberspace, volume 91. Wadsworth Bel-

mont, CA.

Kolb, A. Y. and Kolb, D. A. (2005). The Kolb Learning

Style Inventory - Version 3.1 Technical Speciﬁcations.

Technical report, HayGroup, Boston, USA.

Konert, J., G

obel, S., and Steinmetz, R. (2012). Towards a

Social Game Interaction Taxonomy. In G

obel, S. and

Wiemeyer, J., editors, Proceedings of the Intl. Conf.

on Serious Games (GameDays) in conjunction with

Intl. Conf. on E-Learning and Games (Edutainment),

pages 99–110, Darmstadt, Germany. Springer.

Korossy, K. (1999). Modeling knowledge as competence

and performance. Knowledge spaces: Theories, em-

pirical research, and applications, pages 103–132.

Larusson, J. and Alterman, R. (2009). Wikis to Support the

Collaborative Part of Collaborative Learning. Interna-

tional Journal of Computer-Supported Collaborative

Learning, 4(4):371–402.

Louchart, S. and Aylett, R. (2003). Solving the Narrative

Paradox in VEs–Lessons from RPGs. In Rist, T.,

Aylett, R., Ballin, D., and Rickel, J., editors, Intelli-

gent Virtual Agents, volume 2792 of Lecture Notes in

Computer Science, pages 244–248. Springer Berlin /

Heidelberg.

Mutwarasibo, F. (2013). Promoting University Stu-

dents Collaborative Learning through Instructor-

guided Writing Groups. International Journal of

Higher Education, 2(3):p1.

Onrubia, J. and Engel, A. (2009). Strategies for Collab-

orative Writing and Phases of Knowledge Construc-

tion in CSCL Environments. Computers & Education,

53(4):1256 – 1265.

Rodenberger, C. (2012). Conception and Implementation of

Game Mastering and Team Leadership Components

for a Collaborative 3D Multiplayer Serious Game.

Diploma thesis, Technische Universit

at Darmstadt.

Shell, D. F., Husman, J., Turner, J. E., Cliffel, D. M., Nath,

I., and Sweany, N. (2005). The Impact of Computer

Supported Collaborative Learning Communities on

High School Students’ Knowledge Building, Strategic

Learning, and Perceptions of the Classroom. Journal

of Educational Computing Research, 33(3):327–349.

Smith, A. M., Lewis, C., Hullet, K., and Sullivan, A. (2011).

An inclusive view of player modeling. In Proceedings

of the 6th International Conference on Foundations of

Digital Games, pages 301–303. ACM.

Stahl, G., Koschmann, T., and Suthers, D. (2006). Cam-

bridge Handbook of the Learning Sciences, chapter

Computer-supported Collaborative Learning: An His-

torical Perspective, pages 409–426. Cambridge Uni-

versity Press.

Tychsen, A. (2008). Tales for the Many: Process and Au-

thorial Control in Multi-player Role-Playing Games.

In ICIDS ’08: Proceedings of the 1st Joint Interna-

tional Conference on Interactive Digital Storytelling,

pages 309–320, Berlin, Heidelberg. Springer-Verlag.

Tychsen, A., Hitchens, M., Brolund, T., and Kavakli, M.

(2005). The Game Master. In Proceedings of the sec-

ond Australasian conference on Interactive entertain-

ment, pages 215–222. Creativity & Cognition Studios

Press.

Wendel, V., Babarinow, M., H

orl, T., Kolmogorov, S.,

obel, S., and Steinmetz, R. (2010). Woodment:

Web-Based Collaborative Multiplayer Serious Game,

volume 6250 of Lecture Notes in Computer Science,

pages 68–78. Springer, 1st edition.

Wendel, V., G

obel, S., and Steinmetz, R. (2012a).

Game Mastering in Collaborative Multiplayer Seri-

ous Games. In E-Learning and Games for Training,

Education, Health and Sports - LNCS, volume 7516,

pages 23–34, Darmstadt, Germany. Springer.

Wendel, V., Gutjahr, M., G

obel, S., and Steinmetz, R.

(2012b). Designing Collaborative Multiplayer Serious

Games for Collaborative Learning. In Proceedings of

the CSEDU 2012.

Zea, N. P., S

anchez, J. L. G., Guti

errez, F. L., Cabrera,

M. J., and Paderewski, P. (2009). Design of Educa-

tional Multiplayer Videogames: A Vision From Col-

laborative Learning. Advances in Engineering Soft-

ware, 40(12):1251–1260.

CSEDU2014-6thInternationalConferenceonComputerSupportedEducation