Joris Dormans (2006)
Paper presented at iDiG 2006; the full reference is:
Joris Dormans (2006) "Talking to Games: Gameplay as Expressive Performance" in José Dionísio, António Ramires Fernandes & Paulo Gomes (eds.) Proceedings of the International Digital Games Conference 2006, pp. 133-143.
Abstract. This paper investigates the challenge of designing narrative dialogue in games. This challenge is approached from the perspective that it is mostly a gameplay issue. The notion of gameplay used here stems from an expanded version of the definition of games offered by Katie Salen and Eric Zimmerman. Most notably, this paper adds to their definition the notion of gameplay as interactive player performance and stresses the importance of representation in games. Play is regarded as the production of signs which are evaluated by the game. As such, semiotic notions of signs can be used to help understand the challenge of dialogues in games, especially if we focus less on the signs and more on the way signs are articulated. Strategies for improving game dialogues, which emerges from this perspective, are proposed.
Keywords: Gameplay; Game Design; Narrative Games; Dialogue; Performance.
Many games do a poor job at implementing a system to handle character dialogue and conversation. 'Talking to games' is often an impoverished and repetitive experience. In other words, most dialogues in games constitute rather poor gameplay. In this article I like to examine games and gameplay closely in order to formulate some design strategies that may resolve the challenge posed by dialogues in games.
Gameplay is commonly understood as the player activity: the pressing of buttons on her controller and the cognitive effort that is required to make sense of the riot of sound and vision presented to her by the computer or console. Working with a notion of gameplay that attributes all signifying tasks to the player and reduces her possible interactions to a finite, and generally low, number of buttons, has yielded a particular type of games that reduces player choices to an equally finite and discrete number of options. Especially games that have narrative aspirations suffer from this: structures for interactive narratives often rely on elaborate networks of nodes and connections in which the former represent plot points and the latter the options available to the player. Many people, game scholars and designers alike, have found that stories are not easily translated to, or mediated by, this type of structure. Combining games and stories often results in hybrid artefacts; not really a game and not really a story. Progression through a game can be linked to and mediated as the progression through a story, but this is not the same thing as creating a truly interactive story-game. Designing an interactive game engine for interactive storytelling requires the player to interact with the story while it is being generated. Most games do not allow such interaction, or when they do, allow it only in the most basic and primitive way.
I argue that one of the underlying problems is caused by a lack of understanding of gameplay as expressive performance on the part of the player. Games are generally very good at evaluating player action as low level game performance (does the jump clear the pit, is the gun fired in time to hit an opponent?), but often fail to understand the actions and motivations of the player on a level that would useful for the construction of a meaningful narrative. Although there have been some games that take such aspects of gameplay into account, these attempt are rare and are almost never pursued very far. The method of evaluating gameplay as a form of narrative expression and interaction on the part of the player is currently poorly researched, and I do not expect future research to yield any clear-cut and definite answers soon. Still, I do think that semiotic theory can help to advance games in this area. Semiotics can help to formulate an answer to the question how gameplay can be used to facilitate narrative interaction. In particular, I like to show how a better understanding of gameplay as expressive performance can help us design better gameplay mechanics for the game dialogues required by true narrative games.
The current narrative shortcomings of computer games is no where as evident as in the way typical, contemporary games handle game dialogue. By default dialogue is handled by presenting the player with prescripted text and then offer her a choice between different replies. These choices may, or may not, have a significant impact on the game and its narrative. In some cases choosing between dialogue options is directly translated into different narrative paths for the game to follow. In other cases the options are only superficial, with more, or even all, options leading to the same game result. More often than not the impact of the dialogue is trivial, providing an illusion of interactivity, depth and texture at the narrative level. How elaborate these dialogues are, is largely determined by the production effort and resources invested in them.
The way most games handle dialogue is the direct result of the current state of game technology. Games are played on computers that rely on buttons and joysticks for input. Typing a response on a keyboard is considered too laborious and is only viable for games played on a PC. The artificial intelligent required for reliable speech recognition, composition and synthesises is not available to games, and is unlikely to become available in the near future. What is more, it is generally thought that common gamers are not actors or playwrights and therefore they cannot be expected to produce dialogue of the required level of dramatic and narrative sophistication. Chances are that players will only ruin the narrative experience when allowed to much freedom (Poole 2006: 112-123).
When one investigates game dialogue from a ludologic perspective their poor design is quickly revealed. If a game, as some would have it, is a series of interesting choices, the problem is not lack of choice, but lack of interesting choice. Experienced gamers have become well-versed in assessing which option has what game effect. Most games allow saves to be made just prior to the dialogue, which makes the task of exploring the whole dialogue tree trivial (although sometimes cumbersome). Choice, then, is never made from the perspective of the player's avatar, but always from the perspective of the player and her understanding of possible narrative and ludic consequences. No matter how elaborate and sophisticated the dialogue tree, it is never difficult to digest it entirely and subsequently choose the path that has the desired game effect. It is strange that an industry that makes it his business to confront their customers with a series of interesting choices fails so miserably when trying to use the same medium to convey a conversation.
Today, thinking about games in terms of ludic simulations is fairly well established. That is to say, games simulate real and imagined activities in such way that they provide the player with a fun and challenging experience. To drive a formula one sports car, to borrow a popular example, takes years and years of training in real life, while controlling one in a racing game is far easier and an be learned in the matter of hours, if not less. The racing game tries to evoke the thrilling experience of racing these cars, but without accurately simulating the actual experience. A truly realistic formula one racing game would require the same amount of training, and would quickly seize to be fun. In games we like some sense of realism, but we do not want the real thing it self (Poole 2006: 64) Game simulation is abstracted and fine-tuned to generate a 'fun' or 'stylised' experience (Crawford 1984: 9). What might be realistic is often sacrificed for the sake of gameplay. This perspective leaves us with the question why the experience of being a 'dramatis persona' should be so much more difficult to simulate in a ludic way than it is to simulate a racing car?
Well, one thing that complicates the ludic simulation of a character in an involving story is her performance. A racing car can be easily reduced to an object you can steer left and right, with some added controls for the accelerator and the break, without loss of details important to what you want to simulate. This is then easily mapped to controller of a game console. But how does one map speech, quite important to interactive storytelling, to the same controller? One option is to drastically reduce the number of words you can use. This is the direction Chris Crawford is taking: he has reduced his essential vocabulary for games to under 2,000 words (Crawford 2005: 164-168). But I fear that such a vocabulary can never be reduced to a size that can be conveniently mapped to the buttons of game consoles. A second option is to use only a handful of symbols, map these to controller buttons and use these to create a sense of conversation. This option is, linguistically speaking, more viable than it seems. As Noam Chomsky pointed out, the infinity of expressions in language is created with finite means, and he illustrates that this infinity can be achieved with only a few simple symbols and rules as long as these rules are structured in particular ways (Chomsky 1957: 18-25). A third option is to remove speech under control of the player altogether, but replace it by some gameplay mechanism that is better integrated into the game itself. The way a player acts and reacts can be an important source for dramatic choice and development. Violence, of course, is a very primitive form of communication, one that contemporary video games excel in. The exchange of goods and resources can be taken into consideration or the amount of time spend in particular game-places are more humane alternatives. It is this third option I will focus on and explore in this paper.
Games and gameplay are notoriously hard to define, but an attempt at this endeavour is in order here. And I need to bee more exact than my earlier observation that games are ludic simulations. Katie Salen and Eric Zimmerman, after discussing eight previous definitions of games, define games as "A game is a system in which players engage in an artificial conflict defined by rules, that results in a quantifiable outcome" (Salen & Zimmerman 2005: 80). This definition is constructed around three main elements. (1) Games are systems for artificial conflict defined by rules. This is, more or less, what I mean when I state that games are simulations. Salen and Zimmerman are more precise: they specify that the simulation is a system defined by rules and that it simulates some sort of conflict. (2) There are the players or the participants, without whom the game would not be experienced. And (3) there is a quantifiable outcome: a goal that concludes the game and which determines whether or not the player has won (or in games where a score is kept: how well she did).
These three elements are vital to any definition of games. However, I like to include a fourth element that is left out of Salen and Zimmerman's definition: representation. Games are a form of art, a point made by Greg Costikyan (1994). Games as art involve design of rules but also the design of graphics, sounds and texts. What is more, representation plays an important role in scholarly works of games (Atkins 2003, Carr et. al. 2006, Crawford 2005, Klevjer 2002). It is easy to dismiss this element of representation, but examples such as Gonzalo Frasca's September 12 are clear illustrations of the representational power of games. Games are usually about something, even when their presentation is abstract, such as Tetris:> for even abstraction itself conveys a world-view.
My definition also proposes an altered version the third element: quantifiable outcome. With Salen and Zimmerman this notion downplays the experience and the performance of the player in favour for a 'ludus' style winning, losing and scoring. In my opinion, quantifiable outcome is connected to both the game as a system and to the experience of the player. What I propose is an understanding of quantifiable outcome that can also accommodate 'paidea' play; one that takes into account the performance that preceded the outcome; performance for the sake of performance.
According to Salen and Zimmerman lack of a quantifiable outcome disqualifies role-playing games as 'proper' games (2005: 81). True, there is no real way to win in a role-playing game (even though, in many instances, the death of a player character does signify losing) but that does not exclude the possibility of different resolutions, even when a story develops towards a singular ending. Most role-playing games have a mechanic that quantifies this resolution: the awarding of 'experience points' that can be used to develop your character. These experience points can be rewarded for various feats, from defeating enemies to avoiding traps or solving puzzles. It is not uncommon for pen-and-paper games to reward acting 'in-character' with experience points. In this case the players themselves evaluate the performance of the player and the game offers a system to quantify this as a desirable outcome of play.
Salen and Zimmerman suggest that quantifiable outcome is the end result of playing a game. In that light most role-playing games and other narrative games have no real quantifiable outcome, as these games are either finished or not; there are not many outcomes to quantify. This goes for all games that Jesper Juul (2005: 5) defines as games of progression: games that challenge the player with a series of progressively more difficult tests. The object is to reach the end, and either the player reaches this goal or she gives up (which is not the same as loosing as a quantifiable outcome in, say, a game of chess). Winning is the purpose of playing and eventually ends the game; there is only one outcome. On the other hand, when quantifiable outcome includes an evaluation of the performance of the player then many more potential outcomes enter the equation. It is not only a matter of reaching the end of a game but reaching it in style. For most story-driven games the quality of the experience can be regarded as a quantifiable outcome of the playing of a game. The problem is that such outcomes are much harder to quantify exactly than the binary states of winning and loosing or an ordinal number of points scored.
Many games have a singular end-goal. This goes for most platform, adventure role-playing and first-person shooter games; a significant portion of all games (poor attempts to create multiple plots aside). Many classic arcade games cannot be won and progress towards the inevitable 'game over' screen: Juul uses the eventual demise of the player in Space Invaders to distance the game from its backstory (Juul 2001). At the same time many of these games also have developed strategies to quantify the player's performance. Examples include: points scored, collectable tokens, secret locations, kill-ratio, speed of completion, experience points, and fictional wealth. It is not uncommon for games to combine different strategies in order to cater for different type of players. In some games doing collecting all the tokens and find all the secrets can become an end in itself, while in others players compete to get the best scores, fastest times, or get the weirdest results. Players can be very creative at finding ways to evaluate their performance. So-called sandbox games capitalise on this: there are many ways to create your ideal city in SimCity. Even in games that have a clear quantifiable outcome embedded within its structure the play-performance sometimes matters: one can loose a game but enjoy it at the same time, especially when, despite the loss, the game was well played.
Considering the above, my definition of games expands on the Salen and Zimmerman definition and reads as follows:
A game is an artificial system of rules that defines a conflict and that represents some sort of system, conflict or idea outside the game. Players participate in games and their play-performance results in, or is evaluated as, a quantifiable outcome of the game.
fig. 1. A schematic representation of the four essential elements of games and gaming.
The four elements of this definition: games as rule systems, representation, the player and her performance can also be represented schematically (see figure 1). In this schema the relations between the four elements and the way the form a feedback loop become apparent. Each element consists of a real or physical aspect and an immaterial or psychological aspect. Thus a game is a real system that defines a fictional world. The game world is presented by the production of game text, which consists set of audio-visual signs that can be broken down into their material signifiers and immaterial signifieds. The player herself interprets these signs and forms a mental representation of the represented world. She plays by manipulating a controller which button presses are signs (or at least signals) that are processed and evaluated by the game system. Taken together these signs produced by the player are what constitute the performance of the player.
Of all elements performance is the hardest to observe. The other three elements have a more or less clear physical counterpart. The existence of the player and her role in playing is evident. The game system itself is an object that is produced, even though today it exists mostly as digital data, that data is still physically manifest in the material of DVD disks or in the electric currents on the hardware's electronic circuits. The game text, is far more transient as games produce these texts on the fly. Early scholars of games sometimes mistook the text for the game system and as a result considered only part of what is a game. Whereas the game as a system can be considered, to some extent, as a virtual machine for the production of game texts. Where the game text might still be recorded, performance is even more transient. Although it is possible to film players to get a sense of what it is they do, performance constitutes more than just the pressing of buttons. Each press of the button is evaluated by the game system, and as a result each press becomes a symbolic action within the game world. Often this is directly translated into feedback that is incorporated in the game text, but not always so. Singular actions combine into complex moves and strategies and often it is not just of matter of pressing the right button at the right time, but also a matter of bringing about a state within the game system in which pressing that button is the most effective course of action. Playing is a configurative action, but the result is usually visible for only brief time during a play session, if it is visible at all. The significance of a game action might not even manifest itself outside the context of the game or the mental conception of the player. It is the player's part of the interaction and can be regarded as an expressive activity; as the production of signs and texts that have particular meaning in the context of the game, but which are hard to perceive from outside the game.
Gameplay as an expression, as a form of representation, can be investigated in relation to the three other elements of the schema in figure 1. Gameplay represents actions (or 'moves') within the game system: pressing a button in time equates to defeating a particular obstacle or enemy that stands between the player and her long-term or short-term goal. At the same time the game represents this as an action in a particular fictional world: it matters if the pressing of the button is represented as a jump or a shot; it matters whether we are shooting illegal aliens in outer space or in the suburbs of Los Angeles. Finally, we can consider what these actions say about the player, to what extent these are a form of self-expression of the player.
Gameplay as symbolic action in relation to the representation of a fictional world draws the most attention of these three. Games are notorious for representing violence, raising all matter of objections from worried parents, educators and legislatures. A most striking example of the power of gameplay representation is Raph Koster's hypothetical 'Mass Murder Game' (Koster 2005: 166-171). In this game the player is challenged to throw differently shaped people into a pit. When the people are not spread evenly they may form human ladders and climb out. When they are spread out evenly the people at the bottom are crushed and killed by those on top of them. The premise sounds sinister but the game mechanics are the same as those of Tetris; the Mass Murder Game is nothing more the Tetris in a different guise. Koster's alternative presentation of Tetris stresses the importance of representation in games: there is a world of difference between the methodical, abstract action of stacking blocks and the sadistic (but no less methodical) action suggested by the Mass Murder Game. Although I am inclined to think that the violent representation of most games is not as harmful as popular opinion often frames it to be, the representational power of games should not be underestimated. Critical studies of games in relation to gender issues, for example, do address many problems that games would do well to resolve.
Sandbox games, and other games where the player has many configurative options, are good sources of examples of player expression through gameplay. Many people who play The Sims start experiment with avatars that resemble themselves, their friends and relatives at one point or another. The way a player constructs her empire in Civilization is not only the result of her choices as a player but will likely also reflect her personal opinions of, or experiments with, politics and economy. However, these games respond to players' actions only on the lowest level. Most of these games are not very good at interpreting the players tastes at a higher level, let alone formulate a sophisticated response based on such a higher level understanding. Games that regard and evaluate gameplay actions as expressions by the player are extremely rare, but they do exists. James Newman's discussion of Bladerunner provides us with an interesting example. In this game, like in the film and the book upon which it is based, the resolution of the question whether or not the player-avatar (or main character) himself is an replicant plays a crucial role. In the game this question can be resolved in several different and opposed ways. In some endings the player-avatar turns out to be a replicant, in others he is not. There is not a single moment in which this decision is made, rather, the performance of the player is evaluated throughout the game and this results in the selection of the end that is dramatically fit. The players' response towards other replicants encountered in the game are of particular importance in this evaluation (Newman 2004: 105).
The meaning of gameplay in relation to the game system is understood less well. Gonzalo Frasca, one of the prominent scholars interested in the rhetoric of simulation, observes that our understanding of the representational power of games and simulations is "extremely limited" (Frasca 2003: 225). But even as our understanding in the particular strategies of representation through rules and simulation grow, little has been written about the role of gameplay in this respect. Jesper Juul describes the metaphorical relation between the actions of the player and the in-game actions which to which these refer. But he does not go beyond noticing that in some games there is a similarity in difficulty and in other games there is an arbitrary, metaphorical substitution of in-game actions by a real-life controller action (Juul 2005: 172-173). I think much more can be said about the exact relation between gameplay and in-game actions. My experience with games and the literature on games suggests that clever translation of in-game actions results good or interesting gameplay: see, for example, Steven Poole's account of Shenmue (Poole 2000: 199).
To complicate matters more, gameplay is all these things at the same time. The same button-press that stands for a move within the game, stands for an action that is interpreted in the context of the game text by the audience, and stands for an action that is part of the player's expression. It is a complex representation mapped simultaneously to different dimensions of meaning. These relations of standing for, or at least the first and last instance of this standing for, have in an other context been studied by semiotics. Therefore, it seems reasonable to look at that theory to shed more light on these relations. However, to apply semiotics uncritically would be unwise; we should heed the critiques of many ludologists concerning the application of theories that have developed outside the field of games within the study of games.
An obvious starting point for a semiotics of gameplay is Charles Sander's Peirce's well-known classification of the sign. His categories of icon, index, and symbol classify signs on the relation between the sign and the object it stands for: the icon is a sign that resembles its object, the index is a sign that has a direct existential relation to its object, and the symbol stands for its object by convention (Fiske 1990). It is tempting to interpret Juul's observation of the relationship between in-game and real-life action in this way: distinguishing between a sort of game-play icons (based on resemblance in difficulty) and symbols (arbitrarily, metaphorical relation). However, this would lead to oversimplification, and is in danger of running into the same problems as many attempts to apply semiotics outside the field of linguistics.
Semiotics has a long history of focussing narrowly on symbolic, linguistic signs. Ferdinand de Saussure states that signs consists of signifiers that are arbitrary linked to signifieds and that individual signifiers are distinguished form the signifiers that precede and follow it only by difference in form (Saussure 1983: 100-103). This implies that the human communication works because we learn to recognise and to understand a multitude of signs used in language. For a computer this poses several problems as we still struggle to design good structures that model similar extensive knowledge. Although some these problems can be countered by the smart combination of a relative low number of signs, an alternative solution presents itself in a critique of Saussures statement that "in the language itself, there are only differences" (Saussure 1983: 166).
Saussure uses the word "articulation" to describe the differentiation of signs that is crucial to language. He uses the term in a way that stresses its original Latin root, which means "member, part, or subdivision in a sequence of things" (Saussure 1983: 23). However, in a more contemporary sense of the word, articulation does more than 'make into articles' only: significance may be found in the manner of articulation. In his semiotic study of images and diagrams, Jacques Bertin does exactly this. His theory is based on eight visual variables that are used to differentiate and structure visual signs. Some of these visual variables, such as size are also very good at representing order and ratio in the way they differ. Elements in a diagram can be distinguished because they have different sizes and at the same time this can communicate a particular order – a is larger than b but smaller than c – or a distinct ratio – a is twice as large as b (Bertin 1983: 186-188). A parallel can be drawn the way that poetry, according Roman Jakobson (in Innis 1985: 167-169), makes marked use of speed, rhythm and pacing (the result of a particular manner of articulation) and 'sound symbolism' (a phenomenal connection based on the way words are pronounced). So it may well be that in language itself there is only difference, semiotics often fail to notices that the differences between signs are significantly articulated.
Computers games are able to measure the articulation of gameplay signs quite accurately. The length of time a button is pressed, the exact location of an player-avatar relative to non-player characters, or the timing of an action, are all metrics that contribute to the way signs in a game differ and all are easy for a game to process. A smart game might 'translate' these into significant knowledge about the players identification with particular game characters or whether she is excited, careful or hesitant. This is by no means an easy task, but one that suits computers and games better than parsing meaning from an extended set of arbitrary signs.
When we re-frame the design challenge posed by game dialogue as a challenge to effectively evaluate the performance of the player, a number of potential solutions present themselves along the line of the analysis of gameplay I have presented so far. Strategies for designing game dialogues might include making use of various inputs the computer can easily process and present the result as a dialogue or substituting a dialogue with gameplay system that is a stylised simulation of (some aspects of) conversation.
As stated above, computers have difficulties processing the symbolic signs that make up linguistic communication. A good narrative game requires that the player truly interacts with the story and that the story is embedded within the ludic simulation This does not require the interaction to be linguistic at all (this is where most games go wrong). After all, the gameplay a racing game only has a superficial link to actually racing cars. On similar ground, games could employ many different ways to model or simulate the mechanics of conservation. It may help to present the result of this interaction as proper linguistic discourse, but as The Sims shows even this is not necessary.
One of the most striking features of language and conversation in real life is it richness. Noam Chomsky calls this the discrete infinity of language (Chomsky 1957), whereas John H. Holland uses the term perpetual novelty to describe a similar characteristic of all complex dynamic systems (including language and games; Holland 1998). This is where the dialogue tree fails so miserably, it does not convey the same richness: its mechanics and possible content is too simple, too exposed and too deterministic. Richness can be created by simple (finite) means, as long as these means can be combined in intuitive and non-deterministic ways (Manninen 2003). Games have long been systems par excellence for creating rich, emergent behaviour. It would be strange, and worrisome, if game designers cannot formulate a strategy to convey the richness of conversation within games.
Games as simulations are always subjective and selective, game designers do not model a whole system, rather, they choose some key aspects from the source system and incorporate those in their simulation (Bogost 2006: 98-99; Holland 1998: 24). Most games that include interactive dialogue have chosen to focus on the content of the conversations. But, as we have seen, there is much more to conversation than different words alone. I wish to suggest a gameplay system that focuses less on the 'what' and more on the 'how' of conversation. Less on the different signs but more on the way their difference is articulated.
We use non-verbal communication all the time. This includes body-language, and interpersonal distance among other things. We give goods as token for our affection and say that the real significance is not in the gift but in the act of giving. In his discussion of non-verbal communication John Fiske distinguishes between representational codes and presentational codes. The first focus on the text – on what is communicated. The latter are largely indexical, that is to say, they depend mostly on indexical signs which have a causal relation between the sign it self and what it stands for. With presentational codes there is a direct (causal) relation between what is said and how it is said; there is a direct link between the content and the performance. Most non-verbal codes are presentational, or have strong presentational tendencies (Fiske 1990: 66-67). The best thing about presentational codes is that many are directly measurable and therefore easily processed by computers.
Tony Manninen suggests twelve forms of interaction available in multiplayer games, of which language-bases communication is only one possible form, and of which many can be said to presentational (Manninen 2003). Of these avatar appearance, kinesics, non-verbal audio, spatial behaviour and physical contact appear to be good starting points to start build a new way for games to handle conversation. It is not very difficult to imagine a game that takes into account what a character wears: being heavily armed and armoured communicates something different than wearing an expensive and trendy outfit (avatar appearance). When a player initiates a conversation the distance between the two characters might be an indication of the personal connection between them (spatial behaviour). The way a character moves around a game area – fast and nervous, quiet and observant – might be revealing of the players emotional state and interests (kinesics, spatial behaviour and possibly non-verbal audio). Physically attacking another character is an obvious indication of hostile intent (physical contact).
These things, and others, have been included in games in but rarely in a way that rich behaviour and interaction emerges from the game. Most incorporate these aspects in relative isolation. Games as emergent systems depend largely on simple rules and instances from which patterns emerge which in turn may be used as components in more complex patterns and behaviour (Holland 1998: 142). These systems require that the rules interact with each other. Unfortunately, there have been few games in which the rules governing conversation interact directly with other core aspects of the game. For example, most games have elaborate systems for moving around an environment, almost no game tries to incorporate this system into its mechanics for conversation. This way conversation will always be secondary to the central gameplay, and in many games this only equates to a poor attempt to give the player the illusion of narrative freedom.
Note however, that this type of communication might not be reflected with dialogue or language directly. It might be sufficient for a game to represent the expression of the characters with abstract symbols in the manner of The Sims. In the case of a game it might even be preferable. We do not need a large set of words to create discrete infinity or perpetual novelty. Only when a particular message is to be communicated a game needs to rely on text ("The old man tells you to go to the dungeon and retrieve the sword"). It is also possible to rely on pre-scripted dialogue that serve as a textual intermission or exposition, not unlike the much debated cut-scenes, but which selection is the result of the gameplay. Last, but not least, it might even be feasible to design sub games to handle conversation. After initiating a conversation the games goes into a different mode (much like the way some games handle hostile encounters or inventory sub systems, in fact the familiar dialogue tree is already such a sub game, only with very poor gameplay!) When speaking to another character the player may be required to 'fire' words and phrases at each other, struggling for the control of the dialogue or trying to balance between expressing entertaining nonsense and dry expossion. A sub game where a well 'played' expression might interrupt your partner at a crucial time which leads you to 'win' the debate.
The medium of games is interactive. Although many game scholars acknowledge this, little attention has been paid to interactive performance of the player. The player's actions constitute a gameplay 'text' that is constantly evaluated by the game system. This text is extremely transient, and often directly or indirectly incorporated into game-text. Yet, as a text it is more complicated then the sum of its button presses. Most games evaluate the player's performance only on the lowest level: is the button pressed in time to avoid the obstacle? Whereas a better understanding of play-performance and the way it interacts with the ludic simulation, relates to the game's layer of presentation and it expresses the player long term and short term strategies and motivations, is required to create games that allow interaction on a higher level. But in the end games as emergent systems should be able to excel in this.
The type of game that has the most to gain from the understanding of gameplay as expression is the narrative game. Narrative games have been poorly developed especially because common methods to handle dialogue and conversation, for many a prerequisite for successful storytelling, fail to deliver anything but the most rudimentary interaction. This paper proposes a strategies to improve game dialogues: side-step the problem of dealing with linguistic interaction directly and to simulate it using related aspects of communication. When we make use of gameplay signs which articulation is more easily processed by a computer than the symbolic signs of natural language, communication becomes easier to handle and integrate into the game. We could still present the results of this communication linguistically through pre-scripted (or pre-recorded) texts. When this type of interaction is not central to the core game we better make sure that it constitutes good gameplay nonetheless. It should be not to hard to find ludic system that convey what it feels like to be conversing, which gives the player a sense control without controlling the content of the conversation directly.
With the discussion of gameplay as expressive performance and this strategy, I hope to have shown that much of the depth and breath of the medium of games remains to be explored, and that we are still a long way from mastering the form in all its aspects. I also hope to have inspired designers to take up the challenges posed by game dialogue and to venture deeper into the uncharted waters of narrative games. Just remember, if we want to talk to games, we need to speak their language.
Atkins, B.: More than a Game, The Computer Game as Fictional Form. Manchester University Press, Manchester (2003)
Bertin, J.: Semiology of Graphics. University of Wisconsin Press, Madison (1983)
Bogost, I.: Unit Operations: An Approach to Videogame Criticism. The MIT Press, Cambridge (2006)
Carr, D., Buckingham, D., Burn, A., Schott, G.: Computer Games: Text, Narrative and Play. Polity Press, Cambridge (2006)
Chomsky, N.: Syntactic Structures. Mouton Publishers, The Hague (1957)
Costikyan, G.: I Have No Words & I Must Design. Available at <http://www.costik.com/nowords.html> (1994)
Crawford, C.: The Art of Computer Game Design. Osborne/McGraw-Hill, Berkeley (1984)
Crawford, C.: Chris Crawford on Interactive Storytelling. New Riders, Berkeley (2005)
Fiske, J.: Introduction to Communication Studies, 2nd edition. Routledge, London (1990)
Frasca, G.: Simulation versus Narrative. In: Wolf, M.J.P., Perron, B. (eds.): The Video Game Theory Reader. Routledge, New York (2003) 221-235
Holland, J. H.: Emergence, From Chaos to Order. Oxford University Press, Oxford (1998)
Innis, R.E. (ed.): Semiotics, An Introductory Anthology. Indiana University Press, Bloomington (1985)
Juul, J.: Games Telling Stories? Available at <http://www.gamesudies.org/0101/juul-gts/> (2001)
Juul, J.: Half-Real: Video Games between Real Rules and Fictional Worlds. The MIT Press, Cambridge (2005)
Klevjer, R.: In Defense of Cutscenes. In: Märyä, F. (ed): Proceedings of Computer Games and Digital Cultures Conference. Tampere Univeristy Press, Tampere (2002) 191-202
Koster, R.: A Theory of Fun for Game Design. Paraglyph Press, Scottsdale (2005)
Manninen, T.: Interactive Forms and Communicative Actions in Multiplayer Games. In Game Studies, The International Journal of Computer Game Research 3(1). Available at <http://www.gamestudies.org/0301/manninen/> (2003)
Newman, J.: Videogames. Routledge, London (2004)
Poole, S.: Trigger Happy, The Inner Life of Videogames. Fourth Estate, London (2000)
Salen, K., Zimmerman, E.: Rules of Play, Game Design Fundamentals. The MIT Press, Cambridge (2005)
Saussure, F. de: Course in General Linguistics. Open Court, Chicago (1983)
 In the case of Tetris we can arrive at some interesting interpretations if we are to follow Marshal McLuhan's dictum "the medium is the message". The mathematical consistency and simplicity of the Tetris world combined with the simple, colourful graphics of the interface can be read as a statement. The meaning of Tetris has been the subject of some debates between game scholars. For a summary of this debate see (Bogost 2006: 99-101).