Game Studies

Adapting the Principles of Ludology to the Method of Video Game Content Analysis

Abstract

This article focuses on the methodology of qualitative video game content analysis, aiming to complement the insights provided by Konzack (2002) and Consalvo & Dutton (2006) from a media sociology perspective. A selection was made of 11 contemporary, mature-rated games, and a scheme for analysis was constructed, addressing the different shapes violent activity takes within these games. In a step by step overview of the phases in which the analysis was carried out, the article describes the procedures that were used while constructing a scheme for analysis, while selecting a number of game titles and while training the coders. In the end a brief overview is given of the main results. As such the article provide a blueprint, mapping the difficulties and challenges one encounters while performing a qualitative video game analysis.

Introduction

In recent years a lot of debate has been held on the formal characteristics that mark out the boundaries of video games as a distinct medium, and on the terminology that is the most appropriate for analyzing video games as cybertexts (e.g. Aarseth, 1997), as rule-based systems (e.g. Juul, 2005), as simulations (e.g. Frasca, 2003) or as second-order cybernetic systems (e.g. Kücklich, 2002). Now that the infamous ‘ludology vs. narratology’ dispute has more or less been settled, a consensus appears to exist about the general terminology that should be used when video games are studied, and about the research traditions that could fare well by taking into account concepts such as rule-based play, emergence or simulation in a virtual environment. Nevertheless, very few attempts have been made to operationalize the principles introduced by ludology in a framework suitable for practical, empirical analysis. In this paper such an attempt is made. More specific, an exploration is made of the complexities one encounters when studying the content of a selection of mature-rated video games from a media sociology perspective.

Media effect theory is one of the disciplines where the insights provided by ludology have thus far not been able to conquer solid grounds. Only a few researchers with a background in media sociology or psychology have done the effort of analyzing a number of games, taking into account aspects that go beyond the binary oppositions of ‘violent vs. nonviolent content’ or of ‘sexist vs. nonsexist content’. It is remarkable, on the other hand, that when the effects of playing games on aggression in daily life are debated, aspects such as graphical explicitness, interactivity or realism of the game world (all elements of game content) are to a large degree found responsible for the negative impact video games are assumed to have. Many researchers have associated the mechanisms of script rehearsal and social learning with the interactive experience that is offered in electronic games (Anderson & Dill, 2000), with the increased graphical realism of nowadays games (Sakamoto, 2000) or with the moral justifications that are given for acting violently (Carnagey & Anderson, 2005). Nevertheless, no instruments have yet been developed that allow to investigate or define these constructs in the context of media effect research.

Goldstein (2005:22) brings up this matter as follows: “The context of violent stories … varies along dimensions of realism, involvement, excitement, how violence/conflict begin and end, whether they are presented in an erotic or humorous context. Variations in the nature of video game ‘violence’ have rarely been studied.” In the study reported here, an attempt was to map the variations in the nature of violence in a number of contemporary games. A selection was made of 11 mature-rated games, and a scheme for analysis was developed, inspired by the structural analyses performed by Juul (1999) and Aarseth (1997). However, due to the limited amount of methodological literature available on this type of analysis, a number of methodological difficulties were encountered that had not been accounted for on beforehand. Rather than to provide a comprehensive description of the main results, within this paper it was preferred to adopt a methodological angle. A step by step overview will be given of the different phases in which the investigation was carried out, and of the methodological challenges that were met within each phase. As such it is our goal, not only to make a contribution to the debate on the effects of video game play, but especially, to provide a blueprint for scholars who are interested in performing a systematic video game content analysis.

The paper will be structured in 4 parts. First, a summary will be given of the different perspectives that exist regarding the analysis of video game content, including the point of view defended by ludology, as opposed to the point of view that is preferred within video game effect research. Second, an introduction will be given of the method of qualitative content analysis and of its applicability to the domain of video game studies. Third, the methods that have been used within this study will be described, including the procedures for constructing an analysis scheme, for selecting a number of game titles and for coder training. Finally a brief summary will be given of the main results.

Studying Video Game Content: Different Perspectives

In most theories on the psychological or social effects of video games, only minimal attention is paid to the role of video game content as a moderating variable. When the content of games is studied in an experimental or survey framework, most often this does not go further than to make a basic distinction between violent and nonviolent games (e.g. Ballard & Wiest, 1996; Schutte et al, 1998; Anderson & Dill, 2000; Bushman & Anderson, 2001), or between realistic and nonrealistic games (e.g. Sakamoto, 2000). Different researchers apply different criterions in order to label a game as violent or realistic, and use different games as representatives of the violent or nonviolent category. For example: whereas Anderson & Dill (2000) compare players who have played Myst (nonviolent game) to players who have played Wolfenstein 3D (violent game), Ballard & Wiest (1996) make use of two non-commercial games that were specifically designed for their experiment: a karate game called ‘Karateka’ (violent game) and a puzzle game called ‘Jungle Hunt’ (nonviolent game). As was argued by Sherry (2001) this makes it very hard to compare the results of these studies to one another.

In only a few published studies a systematic analysis has been performed of the content in a selection of popular video games. Dietz (1998) investigated 33 Nintendo and Sega games, with a focus on violence as a predominant narrative, and on the portrayal of female characters. Smith et al. (2003) performed an analysis of 60 popular games from the Sega DreamCast, Nintendo64 and PlayStation consoles, describing a wide range of aspects, such as the amount of violence, the demographics of the perpetrators and targets, the motivations and consequences of violent activity, the means used to perform an act of violence, and the graphical depiction of violent activity. Brand et al. (2003) analyzed 130 games with a focus on the techniques of filmic representation that have been implemented, and concluded that the contents of these video games can be described along a wide range of dimensions, including virtual geography, graphical stylization and narrative structure.

Following Frasca (2003), it can be argued that these studies have investigated elements of representation rather than elements of simulation, and consequently, that a number of characteristics that are essential to the game play experience have been overlooked. In Juul’s definition of video games as formal systems, there is a strong emphasis on the fact that games consist of rules, that an amount of player effort is required in order to overcome the challenges imposed by the game program, and that the outcome of a game is negotiable and quantifiable (Juul, 2005). Juul does not include in his definition the characteristic that video games make use of specific audiovisual codes or narrative techniques. Instead he refers to aspects that are not directly observable in the audiovisual output generated by the game program. Juul acknowledges that there exist lots of games that thrive on graphical detail and on narration, but he adds to this that the use of graphical and narrative codes is neither a necessary nor a sufficient condition for an entertainment product to be referred to as a video game. Studying games, according to Juul, implies interacting with the game rules and exploring the possibilities created by these rules, in addition to studying the graphical codes or the narration that unfolds.

Kücklich (2002) points out that “observing a game necessarily entails influencing it” and argues that a game can only be properly analyzed by means of interacting with its user interface. A game has to be played in order to be understood, and playing a game implies making active choices another player or researcher would not necessarily make. Kücklich claims that there exists no such thing as an ideal player, because it is an essential part of games that players are allowed (and required) to be creative within the framework provided by the game rules. As a consequence it is not possible to grasp the meaning that is formulated in a game, without taking into account specific details of the player context.

Aarseth (2003) makes a similar remark but prefers a more practical solution. He refers to the analysis of Bartle (1996), who identifies four basic types of game players: socializers, whose main enjoyment is the social contact made possible in games; killers, who enjoy hunting down other game characters; achievers, who love competing one another; and explorers, who enjoy discovering different aspects of the virtual game world. Aarseth argues that, ideally, a researcher should play a game several times, each time taking another of these roles, and study what happens in the game as a function of the role that is chosen. Due to time or budget limitations it may not always be possible to play a game for several times. In such cases it will be very important that the researcher develops an awareness of alternative play strategies that could have been chosen by other researchers or players. Aarseth makes a plea for complementing the analysis of a video game with the analysis of secondary resources, such as press reviews or player accounts, as a means to diminish the interpretational bias that is caused by the personal preferences and motivations a researcher necessarily brings along into a game.

In addition to this, Aarseth points out that different players, and consequently, different researchers can become involved in a game at different levels of expertise. He distinguishes four strata of engagement: superficial play, where the researcher plays a game for a few minutes; partial completion, where a few (sub) missions are finished; repeated play and expert play, where a (part of a) game is repeatedly being completed at different levels of difficulty and mastery. Dependent on the specific needs of analysis, each type of engagement can prove useful. Whereas it may be sufficient to play a game superficially when one attempts to make a quick classification of a wide range of games, repeated play or expert play will be needed when one attempts to make a structural analysis of the processes of meaning creation in a specific game or genre.

Method: Qualitative Content Analysis

Given the explorative character of this investigation, it was chosen to perform a qualitative content analysis, taking into account the insights provided by video game effect research, as well as the insights provided by theorists such as Juul, Aarseth or Kücklich. For more than 50 years already, the technique of qualitative content analysis has been considered a valuable alternative for, or addition to the positivist method of quantitative content analysis. In the same year that Bernard Berelson defined content analysis as “a research technique for the objective, systematic, and quantitative description of manifest content of communication” (Berelson, 1952), a different perspective was provided by Siegfried Kracauer, who valued interpretation over objectivity, and argumentation over quantification (Kracauer, 1952). Kracauer argued that, in order to understand the meaning of a media message in its full depth and richness, it is not sufficient to study the manifest content that is communicated. In many cases it may be equally important to investigate the latent meanings of a message: meanings that are not explicitly formulated, but that are implied in the reader’s interpretation. Within Kracauer’s perspective an interpretive role is given to the researcher. As the latent content of a message cannot be analyzed in terms of a strict, quantitative coding scheme, it should be discovered, and its meaning should be explained, described or made plausible, rather than quantified.

When attempts are made to formulate a number of criteria in order to judge the ‘quality’ of a qualitative content analysis (or of qualitative research in general), two main points of view can be observed. Within a first line of thought, the argument is made that the value of a qualitative study should be assessed by applying measures comparable to the notions of validity, reliability or generalizability that are used in quantitative research. Mayring (2000), for example, emphasizes that qualitative content analysis essentially is (and should be) a systematic technique, and distinguishes a number of coding procedures that live up to this requirement. More specific, he identifies three coding techniques that make it possible to systematically extract general tendencies from a text: inductive category formation, explicating content analysis and structuring content analysis. Although these procedures serve different purposes and, accordingly, imply different types of analysis, each of these techniques suggests that the researcher begins with a fixed, predefined analysis scheme, and that this analysis scheme is strictly applied throughout the investigation.

Within a second line of thought, it is also recognized that qualitative research requires quality notions that are similar to those of quantitative research, but an attempt is made to define these notions in a way that is closer to the praxis of doing field research. Barker (2003) stresses the importance of integrating theory and analysis in a meaningful and convincing way, and draws upon a study’s contribution to existing theory in order to judge its value. The relationship of respect a researcher has with the text or respondents that are analyzed is considered an important criterion as well. The argument is made that, while reporting the results of a qualitative study, a researcher should be very open and transparent about the interpretive role that was taken. From a similar point of view, Böhm (2004) describes a number of coding procedures that make use of a predefined analysis scheme, but that, unlike the procedures outlined by Mayring, provide the possibility of changing this scheme based on observations that are made during analysis. Böhn stresses the importance of integrating theory and analysis in a dynamic process that allows to expand, refine or adapt any predefined categories. Following Strauss & Corbin (1990) three types of coding are identified: open coding, where hardly any use is made of predefined categories; axial coding, aiming to complement or refine one or a few general concepts; and selective coding, where the researcher disposes of an elaborated set of concepts that need further specification.

A few authors have made the attempt to apply the principles of qualitative content analysis to the domain of video game studies. Fabricatore et al. (2002) have studied the content of video games from a game design perspective. Combining a grounded theory approach with a user-centered approach that is rooted in usability research, a set of guidelines was elaborated for the development of games that have a high degree of playability. Consalvo & Dutton (2006) elaborated a general-purpose checklist of topics that bear relevance to researchers who are interested in studying video game content beyond the level of representation. Within their scheme the content of an electronic game is split up in the following components: object inventory, interface study, interaction map and gameplay log. Konzack (2002) addresses the issue from a structural point of view and makes a distinction between 7 content categories: hardware, program code, functionality, game play, meaning, referentiality, and socio-culture.

Although within this study a similar effort is made, nevertheless a different perspective was chosen -that of media effect theory- and different methodological choices were made. The procedures that will be described in the remainder of this chapter should not be seen as a form of criticism, but rather, as a complement to the procedures used by the authors mentioned above. In the paragraphs that follow, the methods will be described that were used in 3 subsequent stages of the investigation: constructing a scheme for analysis, selecting games for analysis and training the coders.

Constructing a Scheme for Analysis

In the first stage of the investigation, an analysis scheme was constructed that takes into account representational as well as rule-based elements. Whereas the representational component was modeled on the framework used in the studies by Brand et al. (2003) and Smith et al. (2003), the rule-based component was inspired by the structural models that were developed by Juul (1999) and Aarseth (1997). Juul (1999) identifies 3 constituents of the structure of a video game: the material, containing all textual, graphical and audio files that are included in the game software; the program, consisting of the algorithms that combine the material into an interactive experience; and the output, which is the audiovisual result of the calculations performed within the computer or games console. Aarseth (1997) provides a more detailed elaboration of this general structure, in his model of the internal structure of a generalized, role playing cybertext. Within each of Juul’s components he identifies a number of specific modules, including rules relating to the game world, characters and objects, as well as modules responsible for the audiovisual presentation and the processing of user input.

Based on the authors mentioned above, a general scheme for analysis was developed, structured around 7 topics of interest: audiovisual style, narration, complexity of controls, game goals, character and object structure, balance between user input and pre-programmed rules, and spatial properties of the game world. Each topic is briefly discussed in table 1.

Whereas the methodological purpose of this study was reflected in the identification of a set of general content categories, the theoretical goal of this study was reflected in the specific points of attention that were chosen within these categories. It might be interesting for other researchers to take into account the same or similar content categories (as was done by a.o. Consalvo & Dutton, 2006 or Konzack, 2002), since these are frequently described and discussed in existing video game theory. On the other hand, within these general categories, other researchers might find it more interesting to emphasize other elements of game content, dependent on their specific research needs. Scholars who are investigating the portrayal of gender roles, for example, might rather be interested in the different combat possibilities female and male characters have, or in the degree to which narrative play is included. Similarly, researchers who are interested in the depiction of World War II, might rather be interested in the accuracy with which the weaponry and fighting tactics of that period are simulated, than in the ideology of the reward system.

Selection of Games

After the development of the scheme for analysis was finished, a number of game titles was selected. Two criterions were used in order to include a game in the selection: relevance and diversity. As for relevance, given the focus of this study on violent activity, it was chosen to only include games that have been labeled as appropriate for a mature audience. The ratings given by the American Entertainment Software Ratings Board (ESRB) were used, since at the time of the analysis (April-August 2003) this was the most widely used video game ratings system [1]. It was decided to select a number of games that were dominating the charts of that period (e.g. Grand Theft Auto: Vice City, Max Payne, Halo, Soldier of Fortune II) in addition to a number of classic game titles that had been released a few years earlier (e.g. Half-Life, Carmageddon II). On top of that, three online shooting games were selected: games that are not easily found in the charts because they are often freely distributed (and/or updated), but that are nevertheless very popular among younger players. Since it was one of the main goals of this study to compare the different shapes of violent activity, another criterion for selecting game titles was diversity. Although the games that were selected have all been labeled as ‘violent entertainment suitable for a mature audience’, it was nevertheless chosen to include games that, according to existing typologies and genre classifications, provide a particular type of game play. When making a selection of online shooting games, for example, it was decided to include a tactical game (Urban Terror), as well as an action-centered game (Quake3: Arena). It was also attempted to have a representative of different genres. Not only different narrative themes such as the mafia, science fiction, World War II and terrorism were included, but also different game play mechanisms such as first person shooting, third person shooting, fighting, car driving, narrative play and online competition.

A selection was made of 11 games, having all received a ‘Mature (17+)’ rating of the ESRB ratings board. Three online shooting games were included: Urban Terror (Silicon Ice Development, 2002, PC), Counter-Strike (CS Team, 1999, PC) and Quake III: Arena (Id Software, 1999, PC). Five single player shooting games were analyzed: Half-Life (Valve, 1998, PC), Halo: Combat Evolved (Bungie, 2001, Xbox), Return to Castle Wolfenstein (Activision, 2001, PC), Soldier of Fortune II: Double Helix (Raven Software, 2002, PC) and Max Payne (Remedy, 2001, PlayStation2). In addition, one fighting game was included (Mortal Kombat: Deadly Alliance, Midway, 2002, PlayStation2) as well as one race game (Carmageddon II: Carpocalypse Now, Stainless Software, 1998, PC), and one crossover (Grand Theft Auto: Vice City, Rockstar, 2002, PlayStation2).

The Issue of Irreducibility

The process of selecting game titles did not prove significantly more complicated than the process of selecting television programs or film titles would have been in a classical content analysis. Some more serious difficulties arose when a decision was made about what it practically meant that a game was included in the selection, and would be analyzed according to the analysis scheme that was developed. Unlike films or television programs, where the ‘text’ presented to the spectator is the same every time it is being played, video games have the property of not being reducible to a predefined sequence of signifiers (e.g. Juul, 2005).[2] Every time a (part of a) game is played, the output that appears on the pc or console screen is different from any previous time, even if it is played by the same player under similar circumstances. Therefore, it becomes very difficult to define what belongs to the ‘text’ of a game and what not, and consequently, what will be the boundaries of the object of analysis.

Juul (2005) coins the term irreducibility in order to describe this methodological difficulty: Although the rules of a game system provide a specification of the actions that are possible, the activity that takes place can never be reduced to these rules only. Within the context of this study, a number of practical difficulties were encountered that are -directly or indirectly- related to the characteristic of irreducibility. The most important of these issues are now discussed: game modules; modifications, conversions and add-ons; and hardware-software relationships.

Game Modules

In the software of most contemporary video games, different modules are included, each module offering a more or less radical alteration of the basic game rules. The most common example is the use of different levels of difficulty. Although most single player games make use of a hierarchy that is similar to the basic ‘(very) easy - normal - (very) hard’ pattern, resulting in 3 to 5 levels of difficulty, the differences between these levels are not always elaborated in a similar way. In the sample that was used for this study, in a number of games these differences were mainly a matter of quantity, i.e. of having to face a larger or smaller amount of enemies (e.g. ‘Max Payne’) or of disposing of a larger or smaller amount of ammunition (e.g. ‘Return to Castle Wolfenstein’), whereas in other games the artificial intelligence of the enemies was significantly more advanced at higher levels of difficulty (e.g. ‘Half-Life’). In some games, an extra level of difficulty is included where the player is offered different missions and an extra storyline (e.g. ‘Halo: Combat Evolved’) or different time restrictions (e.g. ‘Max Payne’).

In many of the games that were analyzed not only various levels of difficulty have been included, but also different modes of play. In ‘Grand Theft Auto: Vice City’, players can choose if they want to play in narrative mode, or play a number of missions that are independent of the storyline, such as a Checkpoints Race. In most online shooting games, different competitions can be chosen, such as Deathmatch, Team Deathmatch and Capture the Flag (‘Quake III: Arena’; ‘Urban Terror’) or Bomb/Defuse and Hostage Rescue (‘Counter-Strike’). ‘Carmageddon II’ supports a single player version, where a number of missions have to be accomplished, and in addition allows users to start a network game over a LAN connection. Similarly, ‘Halo: Combat Evolved’ has included a multiplayer mode (local as well as over LAN) in addition to the dominant, narrative play mode.

Modifications, conversions and add-ons

Another recent phenomenon is the inclusion of modifications and add-ons to the source code of a game. There are different gradations to which the source code of a game can be altered, ranging from the use of its graphical engine, over the addition of user-programmed game characters, mission and maps, to the reprogramming of a game’s code in a total conversion. Some games, such as ‘Counter-Strike’, have a history of continuously providing new versions of the same game title, whereas other games, such as ‘Quake III: Arena’ have officially released some of these updates as a separate title (‘Quake III: Team Arena’ being an example of that practice).

The existence of these add-ons makes it very difficult to determine under what conditions a modification should be considered a distinct software product (and thus worthy of a separate analysis), and under what circumstances a modification should be considered a mere variation to the source game (and thus be analyzed as a level or location in an existing game). On top of that, the fact that the most popular online games nowadays dispose of a growing database of user-defined maps makes it practically impossible to make an analysis of the complete ‘text’ of a game title.

Hardware - software relationships

A last aspect that complicates the process of selecting games concerns the fact that the software of a video game is strongly dependent on the hardware it is intended to be played on. Different versions exist of most games, each version designed to be played on a specific platform. In most cases the differences between those versions are a matter of graphical detail (for example, when ‘Halo: Combat Evolved’ was released for the PC it was graphically slightly inferior to the original Xbox version), whereas in some cases there are more radical differences between the release of a game on two platforms. In the 2001 version of ‘Half-Life’, that was released for the PlayStation 2, a few extra maps, missions and weapons were added, that could not be found in the original PC game (which was released in 1998). Related to this is the existence of different versions of a game in different countries or markets. The most noted examples are ‘Caramageddon II’ and ‘Soldier of Fortune II’ that, due to legal problems, have been released in Germany in a modified form (in ‘Carmageddon II’, the color of human blood was changed to green, and in ‘Soldier of Fortune II’ the opponents were changed into non-human creatures).

These three aspects taken together result in the difficulty that it is practically impossible to provide a specification of all features that are part of a game. For reasons of methodological transparency it is very important for a researcher to point out what versions and modules of a title have been included in the analysis, as well as what versions and modules have been excluded. As was demonstrated in this short overview, the differences between separate modules can be limited to graphical detail, but it can also concern aspects such as artificial intelligence, the virtual locations that are used, or the goals a player has to accomplish. Because most of these aspects have been given a central role in the scheme for analysis that was developed, during this phase a number of choices were made that would exercise a large influence in the further course of the investigation. The most important of these choices are now discussed:

Game platform. Of each game not only the title and the name of the developer are specified, but also the year of release and the platform it was released for (see above). In the case of ‘Halo: Combat Evolved’, for example, the Xbox launch game that was released in 2001 was included in the analysis, and not the PC game that was released one year later.

Flemish release. Of all selected games the versions that were released in Flanders (= the Dutch speaking part of Belgium) were analyzed. Unlike other countries such as Germany or Greece, the regulations concerning video game content that are applied in Flanders are not significantly stricter than those applied in most other European countries.

Single player/ multiplayer. As mentioned above, only three games were analyzed as online games: ‘Quake III: Arena’, ‘Counter-Strike’ and ‘Urban Terror’. ‘Mortal Kombat IV’ was analyzed as an offline multiplayer game featuring two competitors. Of the other games, the multiplayer modes have not been included in the analysis. All single player games were analyzed in narrative mode. In the case of ‘Carmageddon II’, for example, it was chosen to include the modules where the player has to accomplish a set of missions, but no attention was paid to the modules where the player can engage in competition with other users or with computer-generated characters.

Missions and maps. In the online shooting games that were selected, it was chosen to restrict the analysis to a number of competitions and maps from the first official release. In the case of ‘Counter-Strike’, it was chosen to have a focus on the maps that provide a Bomb/Defuse or Hostage competition (i.e. maps whose name starts with the ‘de_’ or ‘cs_’ prefix). In ‘Quake III: Arena’ and in ‘Urban Terror’, the analysis was performed on maps that provide a DeathMatch or Capture The Flag competition.

Level of difficulty. In narrative games, it was chosen to restrict the analysis to the second-hardest level of difficulty, because in the hardest level often some a-typical game play elements are included, and because in the less difficult levels often a lot of features are not elaborated with as much detail. ‘Half-Life’, for example, was played at medium level of difficulty, whereas ‘Halo: Combat Evolved’ was played at ‘heroic’ level (the hardest level being called ‘legendary’).

As was the case in the previous step, these choices were directly dictated by the theoretical research goal: to perform an analysis that bears relevance in theory and debate on the effects of playing violent video games. It was not a specific goal to provide a structural analysis of all features that are part of these games, but rather to provide a set of content descriptors that allow making a theoretical distinction between different presentations and simulations of violent behaviour. Therefore, it was chosen to focus on the most common and typical modules, and to attribute only secondary importance to the specific characteristics that are included in separate modes and versions. Once again: other researchers, with different theoretical concerns might have made a different choice in order to overcome this methodological difficulty.

Coder Training and Analysis

In traditional media studies there is a lot of debate on the fact that a researcher necessarily has to make an interpretation of the text that is being analyzed, which has resulted in the development of several research paradigms, such as the etnomethodological approach, the neopositivist approach or a variety of left-critical approaches (e.g. Flick et al., 2004). In the domain of video game studies, this debate has been taken one step further, as the researcher is not only required to make an interpretation of what is visible on the screen, but is also required to make an active contribution to the text that unfolds.

Within this study, it was chosen to follow the guidelines proposed by Aarseth, rather than to perform an additional analysis of different processes of contextualization, as was suggested by Kücklich. The reason for this was mainly practical, and motivated by the fact that the research question was formulated with respect to the fixed rules of the game program, rather than to the discourses that are created during game play. Nevertheless, a number of measures were taken in order to minimize the influence of the researchers’ personal preferences on the analysis and results. The most important of these measures are now briefly discussed.

The sample of games was analyzed within the context of a seminar course on popular culture. 19 students participated in the seminar, each of whom was instructed to analyze one game from the sample. In addition, three games were analyzed by the supervisor of the course. This resulted in a total of 22 game analyses being carried out. Each of the 11 games in the sample was analyzed by two different coders.

In the first weeks of the investigation the coders were demanded to play the selected games for 10 to 15 hours. During this initial period, they were given as few instructions as possible, in order to have them follow their own motivations, intuitions and preferred play tactics. Only after two weeks of playing, they were demanded to reflect on possible alternative strategies, and to consult a number of secondary resources on the game, such as walkthroughs, reviews and discussion boards. They were then instructed to re-play a number of levels, and try to explore different sides of the game, as well as to experiment with alternative styles of playing.

Having had previous experiences with fighting and shooting games was explicitly stated as a requirement for students who wanted to participate in the seminar. As such, most coders were already familiar with the game that had been assigned. Only two students reported having needed several hours of training before they felt capable to play the game they were allocated at a reasonably competitive level. On the other hand, only one coder claimed to have already had hours of experience with his game. Within the terminology of Aarseth (2003), it can be argued that most games were played repeatedly but not by experts.

The coders were instructed to, during the time spent playing, frequently pause the game and to make notes about the play experience they had had. It was demanded to structure these notes around the categories constituting the scheme for analysis. After four weeks of playing and making notes, all coders wrote down a first analysis report, wherein the most important observations were summarized, once again structured along the categories constituting the scheme for analysis. Each coder then presented this report to the other coders (among whom one had been analyzing the same game).

Each presentation was followed by a discussion session, where the other coders could comment on the play tactics that had been used, as well as suggest a number of alternative styles of play. After these discussion sessions were finished, the coders were one last time instructed to re-play the game (or the levels that had already been played) for a few hours, and to take into account the suggestions that had been given by the other coders. Based on the notes that were made during these final play sessions, the initial reports were changed, and the final analysis reports were written.

In short, analyzing the games entailed three stages of playing and making notes. During each stage, the coders were trying to discover different sides of the game program. Within their final report, the coders were demanded to reflect on the differences between the play experience they had had during the first hours of playing (when they were merely following their own preferences) and during the final play sessions (when they were instructed to take into account different possible play styles and tactics). It is remarkable, on the one hand, that only one coder did find it necessary to change the general conclusions of his report, based on the discussion sessions that were held and on the different styles of playing that were used later on. On the other hand, with respect to specific parts of the game program, every coder made some significant changes to the initial report, as a consequence of the observations made during later play sessions. These changes ranged from the identification of more advanced features of enemy artificial intelligence, to the discovery of locations, storylines and characters that had not been met before. One of the coders, who had been playing ‘Soldier of Fortune II’, came to a more detailed analysis of the dominant modes of play in the game, after another coder had commented that in several circumstances it can be more efficient to adopt a patient style of playing, making more efficient use of the game world’s geography, rather than choosing to play the game as energetic and fast as possible.

Results: a Brief Overview

Based on the analyses that were made, a comprehensive and detailed categorization could be made of the different types of violent behaviour that were encountered, similar to the often very complex classifications that have been made by Aarseth (1997), Myers (2003) or Le Diberder & Le Diberder (1993). Unlike the studies mentioned above, the primary goal of this study was not to provide an inventory of the variety of all possible mechanisms of video game play, in order to generate a better understanding of the codes that constitute the video game as a medium. The purpose of this study was to forward a number of content descriptors that can be used in upcoming research on the effects of playing violent video games.

Within each of the categories outlined in the scheme for analysis it was possible to elaborate a number of detailed comparisons between the games that were analyzed. In the context of ‘audiovisual style’ for example, a number of important differences were observed with respect to the level of graphical detail, the use of humor, the film genres that served as a source of inspiration, or the functional use of the soundtrack. Within the category of ‘complexity of controls’, the games were classified based on the use of a 3D navigation system, on the number of moves and functions a player disposes of, and on the feedback that is provided about the use of controls and the goals of the game. Rather than to present the results of the analysis as an extensive enumeration of a wide range of comparisons and classifications, it was chosen to summarize the most important findings based on their relevance in the context of media effect theory. Three topics of relevance have been chosen as a framework around which the presentation of the results is structured: the pace of violent activity, the complexity of violent activity and the reality claim that is made.

Pace of Violent Activity

The pace at which violent activity proceeds plays a significant role in theories on the desensitizing effect of playing violent video games - theories that consider the number of killings adolescents or children participate in while engaging in an electronic game (e.g. Funk et al, 2003). Although the pace of a game, and accordingly, the number of injuries and deaths that occur in a given period of game play, is partly dictated by the strategies a player chooses to adopt, nevertheless certain elements of the programmed game code were found to contribute to it as well. The following content aspects were considered relevant within this respect: virtual locations, artificial intelligence, and narration.

The use of virtual locations. Important differences were observed between the online shooting games ‘Urban Terror’ and ‘Quake III: Arena’. Whereas in the first game, a slower pace of activity is stimulated by the use of realistic environments that offer lots of tactical possibilities, in the second game, a fast, adrenaline-driven play style is stimulated by the use of simple, arena-like geographical conditions. Similar differences were found between, on the one hand, ‘Grand Theft Auto: Vice City’ and ‘Half-Life’ (using a large geographical spread and a lot of variation in locations) and, on the other hand, ‘Mortal Kombat IV’ and ‘Return to Castle Wolfenstein’ (where fast action is stimulated by the inclusion of few tactical possibilities in the world geography).

The artificial intelligence of NPCs. More recent game titles in the selection, such as ‘Halo: Combat Evolved’, were found to stimulate a more cautious style of playing, because enemy characters are included whose behaviour cannot be easily predicted, as well as allied characters that can be assigned a wide range of tasks and functions. In this respect, multiplayer games were generally found to require more caution and anticipation than single player games.

Narration. Within games that thrive on a strict and linear narration (e.g. ‘Max Payne’) or within games that require a lot of exploring (e.g. ‘Grand Theft Auto: Vice City’ or ‘Half-Life’), the action sequences are frequently interrupted with cut scenes or side missions like solving a puzzle. Compared to pure action games like ‘Mortal Kombat’ or ‘Counter Strike’, acts of shooting, beating or driving over an opponent succeed each other at a lower pace, leading to a smaller number of deaths and injuries over a given period of time.

Complexity of violent activity

The complexity of the action that takes place within a game is forwarded as an important variable in a number of studies that associate video game play with educational benefits such as goal-oriented thinking (Blumberg, 1998), spatial cognition (McClurg & Chaillé, 1987), memory rehearsal (Oyen & Bebko, 1996) or eye-hand coordination (De Aguilera & Méndiz, 2003). Within our sample of games, at least two content elements were identified as contributors to the cognitive effort that is required of the player:

The complexity of the controls. Within several games the player is offered a wide range of possible moves and instructions, which results in a relatively long initial period of practicing and training before one has reached a minimal amount of skill or expertise. For example, the combination of basic moves into combos that is part of ‘Mortal Kombat IV’, the messaging facilities included in ‘Counter Strike’ or ‘Quake III: Arena’ or the possibilities for keyboard customization elaborated in ‘Soldier of Fortune II’, were found to increase the cognitive effort that is required of a player.

The complexity of the character and object systems. Whereas the developers of ‘Counter-Strike’ (with the use of an extensive trade system) or ‘Mortal Kombat IV’ (with the inclusion of a wide range of fighter characters, each disposing of specific skills and weaknesses) have stimulated the player to perform a lot of on beforehand strategic thought, the designers of other games (most notably, ‘Carmageddon II’, ‘Grand Theft Auto: Vice City’ or ‘Max Payne’) have attempted to minimize the number of aspects that should be taken into account while selecting or equipping an in-game character.

Reality claim and moral justifications

In most discussions on the hazardous effects of playing violent games, specific game titles have been criticized for morally rewarding and stimulating the use of violence in a realistic environment (e.g. Sakamoto, 1994; Anderson & Dill, 2000). With respect to both realism and the justifications that are given for performing an act of violence, a number of important differences were observed within our sample of games. In the context of the reality claim that is made, a number of aspects were identified as inhibitors of the verisimilitude of a game title, including humor (e.g. ‘Return to Castle Wolfenstein’, ‘Grand Theft Auto: Vice City’), defying the laws of physical movement in order to create spectacular possibilities for action (e.g. ‘Mortal Kombat IV’, ‘Carmageddon II’), or the use of extravagant color effects and light experiments (e.g. ‘Halo: Combat Evolved’, ‘Half-Life’). In the context of the justifications that are provided for violent behaviour, differences were observed between, on the one hand, games that maintain a strict division between right and wrong (e.g. ‘Max Payne’) and, on the other hand, games that stimulate the use of violence for personal motives (e.g. ‘Soldier of Fortune II’) or for no apparent reason at all (e.g. ‘Carmageddon II’).

Conclusions

The analysis of the programmed text of a selection of mature-rated video games according to the principles of ludology brought along a number of methodological difficulties that have not been documented in traditional works on content analysis or text analysis (e.g. Titscher et al., 2000). It was therefore chosen to structure this paper around the different phases in which the analysis took place, and to provide a detailed description of the most important methodological difficulties that were encountered within each phase. In general, it was found that traditional methods of content analysis face three types of challenges in the light of the concepts that have been elaborated by ludology:

1. To develop an analysis scheme that takes into account aspects of simulation in addition to aspects of representation;
2. To mark out the boundaries of what belongs to the text of a video game
3. To deal with the fact that analyzing a game entails playing a game, and consequently, that the object of analysis is partly a construction of the motivations and preferences a researcher carries along into the play experience.

In addition, presenting the results in a way that is constructive to the debate on the effects of playing violent video games brought along one extra issue:

4. To extract a number of content descriptors that can be used in upcoming research on the effects of playing electronic games.

Within the methodology of qualitative content analysis, aspects such as interpretive transparency, providing a convincing report and closely integrating theory and analysis, have always taken an important place. Different perspectives exist on how to define concepts such as validity, reliability or generalizability in the context of qualitative research. A recurring point of discussion is the observation that a researcher necessarily makes an interpretation of the texts that are investigated, and as a consequence, that a qualitative analysis is necessarily performed in a more or less biased way. Some authors have attempted to limit a researcher’s freedom in being constructive of the meanings that are created, and have proposed the use of strict coding procedures. Others have highlighted the fact that interpretation often yields a wealth of rich information, and have made a plea for a meaningful integration of analysis procedures and theoretical perspectives.

Based upon the methods that were used within this study, it can be concluded that, within the context of video game studies, the issue of the interpretative position of the researcher is even more relevant than it already was in the context of traditional text analysis. Within interactive texts such as electronic games, a researcher not only makes an interpretation of the audiovisual output that appears on the computer or console screen, but also contributes actively to the messages that are conveyed. As was described within this chapter, this characteristic brings along a number of practical challenges, and forces a researcher to make a number of far-reaching research decisions.

The decisions that were made within the framework of this study have been directly inspired by our theoretical concern: to map the different shapes violent activity takes within the selected games. Within each step it was explicitly pointed out that researchers with other theoretical concerns might have made different decisions, dependent of the specific needs of the study that is performed. It would be beyond the scope of this article, and it would even testify of short-sightedness, if we would conclude this article by formulating a number of specific guidelines on how a qualitative video game analysis should be carried out, based upon the methods that were used within this study. Rather than that, we hope to have raised the awareness that, as video game researchers we are intensively involved within the texts we analyze. It is extremely important that we do not take for granted our own motivations, level of expertise or content preferences. Instead the video game content researcher should develop a strong sense of self-awareness, and acknowledge that many methodological decisions must be made during the course of an investigation. As was pointed out within this article, these decisions are a direct function of the theoretical research goals, and accordingly, of relationship we choose to maintain with the games we are analyzing.

Endnotes

1 The Pan European Game Information (PEGI) system we know today had not yet been founded. The predecessor of the PEGI system, the ratings system of the Entertainment and Leisure Software Publishers Association (ELSPA) was the most widespread system in Europe, but several countries, including Germany and France made use of a system of their own.

2 Even though the definition of a text as a ‘communicative event’, which has broadened the notion of a text to a variety of everyday signs and practices, has become widely accepted in recent years (e.g. Titscher et al, 2000), within this paper it was chosen to restrict the notion of a video game ‘text’ to the programmed functions and data structures that constitute the software of a game. As such the different practices and player discourses that are created during game play are excluded from our definition. The reason for this is mainly practical: maintaining a strict distinction between game program and gaming practices made it possible to provide a clear description of several methodological difficulties that were encountered, as well as of the specific solutions that were elaborated within this study.

References

Aarseth, E. (1997). Cybertext: Perspectives on ergodic literature. London: The Johns Hopkins University Press.

Aarseth, E. (2003). Playing Research: Methodological approaches to game analysis. Melbourne: Proceedings of the 5th digital arts & culture conference.

Anderson, C.A. & Dill, K.E. (2000). Video Games and Aggressive Thoughts, Feelings, and Behaviour in the Laboratory and in Life. Journal of Personality and Social Psychology, 78(4), pp.772-790.

Ballard, M.E. & Wiest, J.R. (1996). Mortal Kombat ™: The effects of violent videogame play on males' hostility and cardiovascular responding. Journal of Applied Social Psychology, 26(8), pp.717-730.

Barker, M. (2003). Assessing the 'Quality' in Qualitattive Research: The Case of Text-Audience Relations. European Journal of Communication, 18(3), 315-335.

Bartle, R. (1996). Hearts, Clubs, Diamonds Spades: Players who suit Muds. Retrieved 07/08/07 at www.mud.co.uk/richard/hcds.htm

Berelson, B.(1952). Content Analysis in Communication Research. New York: Free Press.

Böhm, A. (2004). Theoretical Coding: Text Analysis in Grounded Theory. In U. Flick, E. von Kardorff, and I. Steinke (eds.), A Companion to Qualitative Research. London: Sage.

Brand, J.E., Knight, S. & Majewski, J. (2003). The Diverse Worlds of Computer Games: A Content Analysis of Spaces, Populations, Styles and Narratives. In M. Copier. & J. Raessens (Eds.), Level Up: Digital Games Research Conference. Utrecht: Digital Games Reasearch Association.

Bushman, B.J. & Anderson, C.A. (2001). Violent video games and hostile expectations: a test of the general aggression model. Personality and Social Psychology Bulletin, 28(12), pp.1679-1686.

Carnagey, N.L., & Anderson, C.A. (2005). The effects of reward and punishment in violent video games on aggressive affect, cognition, and behaviour. Psychological Science, 16, 882-889.

Consalvo, Mia & Dutton, Nathan. (2006). Game analysis: Developing a methodological toolkit for the qualitative study of games. Game Studies 6 (1).

Dietz, T. L. (1998). An examination of violence and gender role portrayals in video games: Implications for gender socialization and aggressive behaviour. Sex Roles, 38, 425-442.

Edwards, R. (2001). GNS and other matters of role-playing theory. Adept press. Retrieved 07/08/07 at http://www.indie-rpgs.com/articles/1/

Fabricatore, Carlo, Nussbaum, Miguel & Rosas, Ricardo. (2002). Playability in action videogames: A qualitative design model. Human-Computer Interaction, 17 (4), pp. 311-368.

Flick, U., von Kardorff, E., and Steinke, I (2004). What is qualitative research? In U. Flick, E. von Kardorff, and I. Steinke (eds.), A Companion to Qualitative Research. London: Sage.

Frasca, G. (2003). Simulation versus Narrative: Introduction to Ludology. In M.J.P. Wolf & B. Perron (Eds.), /Video/Game/Theory/. London: Routledge.

Funk, J.B.; Buchman, D. (1996). Playing violent video and computer games and adolescent self-perception. Journal of Communication, 46, 19-32.

Funk, J. B., Buchman, D. D., Jenks, J., Bechtoldt, H. (2003). Playing violent video games, desensitization, and moral evaluation in children. Journal of Applied Developmental Psychology, 24, 413-436.

Goldstein, J. (2005). Violent Video Games. In J. Raessens & J. Goldstein. (eds.) Handbook of Computer Game Studies (pp. 341-358). Massachusetts: The MIT Press.

Johnson, S. (2002). Emergence : the connected lives of ants, brains, cities and software. London: Penguin.

Johnson, S. (2005). Everything Bad is Good for You. London: Penguin.

Juul, J. (1999). A clash between game and narrative. M.A. Thesis. Retrieved 07/08/07 at http://www.jesperjuul.dk/thesis

Juul, J. (2005) Half-Real.Video games between rules and fictional worlds. Massachusets: The MIT Press.

Konzack, L. (2002). Computer game criticism: A Method for computer game analysis. Tampere, Finland: Proceedings of the Computer Games and Digital Culture Conference.

Kracauer, S. (1952). The challenge of qualitative content analysis. Public Opinion Quarterly, 16, 631-642.

Kücklich, J. (2002). The Study of Computer Games as a Second-Order Cybernetic System. In F. Mäyrä, (ed), Computer Games and Digital Cultures (pp101-111). Tampere University Press.

Kücklich, J. (2003). Perspectives of Computer Game Philology. Game Studies, 3(1).

Le Diberder A. & Le Diberder, F. (1993). Qui a peur des jeux vidéo? Paris: Éditions La Découverte.

Lee, K. M. (2004) Presence, Explicated. Communication Theory, 14(1), pp. 27-50.

Mayring, P. (2000). Qualiative Content Analysis. Forum: Qualitative Research 1 (2), accessed 07/08/07 at www.qualitative-research.net/fqs-texte/2-00/2-00mayring-e.htm

McClurg, P. & Chaillé, C. (1987). Computer Games: Environments for Developing Spatial Cognition? Journal of Educational Computing Research 3(1), 95-111.

Myers, D. (1992). Time, Symbol Transformations, and Computer Games. Play & Culture, 5 (4), pp.441-457.

Myers, D. (2003). The Nature of Computer Games: Play As Semiosis. Peter Lang Pub.

Oyen, A. & Bebco, J.M. (1996). The Effects of Computer Games and Lesson Contexts on Children’s Mnemonic Strategies. Journal of Experimental Child Psychology, 62, 173-189.

Sakamoto, A., (1994). Video game use and the development of sociocognitive abilities in children: Three surveys of elementary school students. Journal of Applied Social Psychology, 24(1), pp.21-42

Schutte, N.S., Malouff, J.M., Post-Gorden, J.C. & Rodasta, A.L. (1988). Effects of playing videogames on children's aggressive and other behaviours. Journal of Applied Social Psychology, 18(5), pp.454-460.

Sherry, J.L. (2001). The effects of violent video games on aggression: A meta-analysis. Human Communication Research, 27(3), 409-431.

Smith, S. L., Lachlan, K. & Tamborini, R. (2003). Popular Video Games: Quantifying the presentation of Violence and Its Context. Journal of Broadcasting & Electronic Media, 47(1), pp.58-76.

Strauss, A. and Corbin, J. (1990). Basics of qualitative research: Grounded theory procedures and techniques. Sage Publications.

Titscher, S. Meyer, M., Wodak, R. & Vetter, E. (2000). Methods of Text and Discourse Analysis. London: Sage.