Log in


A Narrative Discourse Analysis of Closed Room Mysteries

Recent Entries · Archive · Friends · Profile

* * *
Recently we've been seeing a revival of the mystery genre known as 'the closed room' problem. First used in a novel called The Yellow Room, such stories revolve around several guests, who don't know each other, slowly being murdered, lots of possible solutions and most importantly: no way out. Even though we could argue that it all started with The Yellow Room, we should credit Agatha Christie with constructing the template for most if not all of the works of this genre. Christie added two important twists to the concept of a murder that happens in a place where there is no apparent ingress and egress and where there are a well defined set of suspects. What Christie did was she kept the guests prisoner in the 'yellow room' for the duration of the mystery and those guests or suspects would be killed one by one until no one would be left. So how does this form of narrative work and why is it interesting to know how it works?

Before you read this any further let me warn you. In order to explain the narrative patterns and constructs I'm going to have to reveal many well known plots. If you do not want to risk finding out the dunnit and the who, then stop reading.

Before discussing what we can do with narrative knowledge about closed room mysteries, let's first look at what we're up against. If we're talking about the kinds of mystery novels or movies that revolve around this closed room dilemma, then In all likelihood you will read a description that can be summarized as: two or more strangers are invited to a mansion/island/house/room by a host none of them knows. As they slowly become acquainted they also slowly die one by one until there is only one or no one left. The premise never changes except for some of the details of the setup and the conclusion. For example in the movie Cube we, as well as the trapped strangers, are immediately confronted with the fact that none of them knows what they are doing in a small room or how they got there. There is no lead-up where people explain to one another how they were all invited to this curious party at some remote location. A rather popular version of this same plot device can be seen in the movie Saw, the other plot mechanisms are however exactly the same. The other way in which closed room mysteries can differ amongst themselves is how they end, which can be one of two ways. Either one of the guests escapes by figuring out what has been going on all along, such as in the movies Exam, Saw, The Method and Fermat's Room or none of the innocent guests survive such as in Devil, which is the most common variety and the one Agatha Christie started with in Ten Little Indians. I would like to leave out such films as House on Haunted Hill, The Legend of Hell House and The Thing, etc, because they use either fantasy or supernatural elements making the narrative unreliable since anything can happen and any imaginary entity can intrude at any time.

This last detail, the one where nobody survives the ordeal, is actually a rather big clue because then the killer somehow must have thought to have died. Other than the movies Cube and House of 9, all other closed room plots have the killer or planner be part of the crowd that is trying to figure out the puzzle. A quick note on the film House of 9. It is in a lot of ways the odd one out. After watching this production, which has a delightful twist at the end, you realize the goal of the film isn't puzzling or problem solving. Instead we have to conclude that the director wanted to explore clashing personalities. There are two strong pieces of evidence noted by a lot of other reviewers. First of all those trapped do not make a concerted effort to find out if there are truly no exits. Also the writer of the script wonderfully decided to play out unlikable characters exactly as they would behave in real life as opposed to film where they would be arrested, imprisoned or killed. Another important detail to note is how a movie like The Method throws out a lot of important principles of mystery writing. It is therefore difficult to say something sensible about how the film fits into the format we've defined. For example at some point as the hopeful job candidates (read victims) are given some breathing space and time to relax because lunch is provided, a seemingly important scene plays out. To everyone except one of the known insiders, or informers, the lunch tastes bad and is possibly even spoiled. All job candidates pretend eventually that there's nothing wrong with the food and force eat it anyway. Except one of the characters. To anyone observing this sequence of events and character behaviors the conclusion would be that the individual making an opposite decision from the group must have some significant meaning. As it turns out, it doesn't. This is a clear instance of a writer not understanding Tchekov's Gun.

With all of these pieces in mind we can now say something about why stories like this are interesting.

The Closed Room Narrative

We have a group of people that never leave a space, who have to interact to survive and who will have very different personalities. As you can see, the restricted nature of the setup gives us a unique opportunity to study narrative. Because all events take place in a small confined never changing arena we can discard any influence on the story based on location. Characters aren't running around a city or urban area and we can therefor establish very exact times and places. As in: places within a room or rooms. Besides location we can also take out hidden events since everything occurs out in the open right in front of our eyes. There are no situations where we do not know what one of the characters was doing because all of them are always in view. Even when people are out of sight we are usually informed of their doings. In fact under the 'rules of mystery writing' we have to know what suspects have done otherwise the reader can't puzzle along.

With all those extraneous details out of the picture we end up with a sequence of events that truly matter. As in: we can now better judge how each event contributes to the puzzle. Normally when we read a novel or watch a movie we are treated to lots of extra details and character developments. Some of these are important to the narrative of the mystery and some of them are only important to the characters in the story. In a closed room mystery we can look at every event in light of the puzzle and the narrative. Some of them might turn out to be unimportant but in those cases the events turn out to be red herrings. We know this is the case for a number of reasons. First of all when you want to kill off eight to twelve characters in the span of an hour and a half you will need to have every event count. A second reason why we can now take events at face value is that we can better tell which events are real-time and which ones are reflective. Reflective means that events are described or mentioned by the characters in the room but are not actually happening at that point in time in the room. That means we can carefully separate what happened to whom and when. Characters might still be lying about what happened but at least we know more about the where and when.

We can now start to ask our main questions about narratives using the closed room experimental setup: what do we think is happening in the story v.s. what is actually happening? In more accurate narrative terms: how consistent are the perspectives (understanding of what is happening) and positions (level of opportunity to change events) of the characters compared to our own perspective. Are those two always different? Or are those two always the same perhaps? Quit frequently I have to or want to explain the difference between narrative and story. It sounds trivial and pedantic but there is an important difference. Understanding the difference can even help you enjoy a novel or movie more. Here's why. I will be using E.M. Forster's basic explanations of narrative, plot and story [1] to begin with:

A narrative is simply a sequence of events. If you were to write a novel that only has narrative it would look like: "and then ... and then ... and then ...". Narratives of this sort are quite boring because it doesn't tell us anything about people. Ironically even though the main construction mechanism is the "and then ..." structure it doesn't tease us to ask: and then what happened? In order to accomplish that elusive and important element "and then?" we have to turn the narrative into a story. A story starts to emerge when the events in a narrative have a significance and meaning to the characters they are about. More so when those events drastically changes their experience or even outlook on life. Things get interesting when, we, as readers, start to detect patterns between the narrative (events) and the characters' behavior in a story. That's what we call plot. A plot describes the causality between people's behavior and events, or between events in general. In a dryer form could even say that a plot consists of a causal chain which might only be apparent at the end of the chain of events.

Discourse Analysis

A plot, perhaps better called a character causality puzzle, is what we're essentially dealing with here when we look at Closed Room Mysteries. It's tempting to immediately delve into the technical details of event causality, but if we did that we might miss a lot of the human elements. Instead it makes more sense to use a method like Discourse Analysis [2], which can capture different perspectives and interpretations of a text. In Discourse a number of participants who are communication use various techniques to create a persistent and consistent reality. We will use the word 'means' for the methods characters use and 'creation' for the things they build or construct. So what do we mean by 'construct'? When people try to explain a situation to others and to themselves they construct a plausible explanation or description. That explanation is such a construction. But it is more than a formal way of saying what has just happened. It is also a self-contained reality with its own rules and standards.

This is pretty abstract stuff so let's take a look at an example first. Let's say that five people are for some mysterious reason trapped in a room which has no discernable openings, no means for anything to come in and no way for anything to get out. Most likely the question which will go through everybody's mind is: "how did we get here?" Depending on the roles and personalities one or more of the 'victim's will offer an explanation. This explanation might clash with the beliefs and sensibilities of the other subjects and they might step up and question the reasoning. What ensues is a back and forth between the trapped characters in an attempt to make sense of their predicament. They are effectively co-constructing a way of thinking about their situation. That's quite complicated because it would have been easier to say that they've come of with a rationale, but that wouldn't help them later on. What they need is a system of thinking that can help them in their overall situation. This way of thinking or reasoning can be considered the creation in terms of Discourse Analysis. When we apply Gee's general discourse theory [2] we should look at 7 versions or perspectives on reality. These perspectives are called:
  • Significance
  • Practices
  • Identities
  • Relationships
  • Politics
  • Connections
  • Sign systems and knowledge
Think of these perspectives as ideal states of understanding given a certain context or situation, or a lens that provides clarity and filters the characters' understanding. Our trapped victims, let's call them subjects instead, all together try to generate consistent versions of Significance, Identities, Relationships, etc. For example, our poor characters might be trying to create Connections. Either connections between themselves to figure out if perhaps they have something in common that brought them there and that could get them out of their predicament, or connections between events that happened to perhaps find a link or reason that explains what is going on. Think of the dinner scene in the movie Clue where each of the guests realizes they have some connection with the political world of Washington. The exact same scene occurs in Murder by Death where ironically all the guests are famous murder mystery writers. Another example would be that throughout the interaction between characters certain aspects of Significance are created. Think of those as highlights in the experiences of all characters/participants.

We need to make this one step more complicated because in narrative, particularly this kind, we need to add the narrator/author as a character who creates Connections and Significance, etc, as well. Now here we have to be extremely careful because the narrator and the author (and even the person telling the story) are not the same person, as it was very clearly explained by Ball [4]. Essentially the narrator's job is to make the narrative as clear and as accessible as possible to the audience. The author might have different goals, most of which are usually personal. Some authors use a story to analyze their own past, some want to make a viewpoint come across or deliver a message. Whatever the reason, since the author can have an unduly effect on our analysis, due to this intruding, external and irrelevant perspective as it were, we will add only the narrator as an additional character to our Discourse Analysis and we will treat this narrator as if he or she plays an active part, usually the part of hiding important information such that the subjects have to work rather hard to either make sense of their predicament or survive.

To accomplish this, to build the 7 versions of perspective, they have six methods to their disposal. Again this is according to Gee's theory, but we will see that it works out quite well for closed room mysteries and it can even help us out to understand what the dynamics are and narrative is in plots like these. In short our characters have the following means to make sense of their narrow world:
  • Social languages
  • Socially Situated Identities
  • Discourses
  • Conversations
  • Figured worlds
  • Intertextuality
A full explanation of all these methods and resulting creations is beyond the scope of this article, for more information please consult Gee's work on Discourse Analysis [2]. Instead of looking at all combinations of means and result for a full analysis instead we could look at those combinations especially vital for closed room narratives. Granted, all the combinations are potentially valid but some are more active in the foreground than others. What is meant by that is that some of the combinations are especially important for closed room mysteries.

Significance and Connections

Let's begin with the big stuff, the obvious stumbling blocks our subjects have to deal with. They do not know where they are and they do not know how they are related to each other or to the events that brought them there. In essence they are trying to find links and relationships. Most of the time the subjects try to find bits of information in the environment and the other subjects that stands out, something that is out of the ordinary. A ridiculous notion you might think because their very situation is extraordinary. Still, humans are extremely fast at adapting to novel situations and within minutes (film and novel minutes) they have accepted that they are in bizarre circumstances. But something else has to stand out, they are ordinary people who have no business being there, at least that is most likely their starter set of thoughts. How do they figure out what is now Significant in this new environment? What are the new rules of existence? Without anything to go on the subjects have to either try things out or through Conversations with the other subjects gather more details.

An interesting phenomenon plays out in most of the movies and novels listed, the subjects try to determine who is a friend and who is a foe. They go through this process even though there is no immediate evidence that any of the other subjects is the enemy in the story. For a lot of the stories presented here an adversarial situation is created whereby subjects have to compete for survival, but that doesn't automatically make a subject The unseen enemy. Instead we should think of this as a means or a technique which we can call the Safety Discourse. Discourse with a capital D as it is one of the seven means listed above. Humans, and also animals, will want to get an immediate sense of who can help us and who will hinder us or endanger us before we do anything else. In fiction the author has the characters use Social Languages and Socially Situated Identities to figure out who is what. Think of it this way, a man all of a sudden walks into our subjects' prison wearing a police uniform and is carrying a flashlight, this is our accepted way of signalling that he is one of the good guys. Of course this could be completely misleading and there are a number of examples where authors have used this accepted Socially Situated Identity to fool us. In our definition of a Closed Room Mystery there is nobody coming in from the outside and the characters will have to determine by picking up subtle clues what the collection of Identities are. Usually there are a couple of tropes we can count on. There is always the leader character, the paranoid character, the wise character (tends to survive) the rash character, etc. Whatever combination of Socially Situated Identities are being signaled we can count on them being clearly identified to us.

Now that our subjects have figured out what roles are amongst them they have at least taken away some of the insecurity of their predicament, they now have something they can count on. Or rather: the author of the narrative now has a means of making the events move forward with some basic assumptions. Because, really, the role assignments are there for us so that we can be funneled towards a conclusion or a point in the narrative where we can see the story as a whole. See Narrative Concepts for AI Driven Digital Interactive Story Telling [3] for a discussion about the difference between a conclusion or turning point versus a reveal.

Sign Systems and Knowledge

You might have gotten a sense that so far the Discourse Analysis describes some sort of negotiation, one between you the reader/observer and the author, accomplished by the proxies of the narrator vs the characters. Thinking about plot driven narrative as a negotiation is not a bad analogy. The author through the narrator gives the characters hints, leads, information and experiences and we're picking and choosing and feeling our way along. We might agree with the narrator (author) or we might disagree, but we know what we're both talking about. In other words we've built up a mutually agreed set of rules by which the small confined world works. You could think of this shared narrative model as a form of Sign System combined with context specific Knowledge.

How do subjects communicate and test the newly discovered rules of their demise? Most often they use Figured Worlds and Intertextuality. When you use a figured world you will couch a conversation in terms borrowed from well known contexts. The simplest example would be someone who uses baseball terms metaphors consistently in daily conversation. Think of it as speaking in metaphors borrowed from well worn tropes. We've already seen one such trope, which is the tendency to think in terms of human hierarchies. One of the subjects should be the leader, one should be the executive branch who gets sent out to do things and examine places, etc. Agatha Christie ironically had her characters always drawn from a static pool of roles. We therefor end up with characters whose role in the narrative needs no further explanation. For example in the story 'And Then There Were None', the leader is a figure whose profession is (was) a judge. The executive branch is represented by a police detective together with a retired colonel. Christie always deployed these types of characters because she could then focus on the puzzle not the participants. Other stories are not so clear about role divisions, which can greatly enhance the depth of a narrative or it can greatly confuse things if there are no other models the reader/observer can depend on. For example in the movie House of 9 there are no leaders and no followers. Even the notion of aggressor vs victim doesn't quite exist. Unfortunately there are no other support systems for the viewer to get a good sense of what is really going on and the film feels very scattered and unstructured. Nothing wrong with that except the very premise of the story is that there is a logical plot in which characters exist, instead of a number of characters that try to figure out if there is a plot at work.

Figured Worlds are very powerful mechanisms for getting a point across. It accesses common knowledge to rapidly build an understanding of a complex situation. But most of it is unspoken and assumed, we can't say much about it and it would be difficult to mimic in a digital system since it requires constant checking if the live participants are on board with the proposed model. A more direct approach is the use of Intertextuality since it is a direct quotation of references from external models. If we go back to the baseball example we could say that the actual quotes produced by our neurotic baseball obsessed subject are what Intertextuality is about. It appears that Intertextuality is a more overt form of Figured Worlds. This is in essence true but when using Figured Worlds we draw various elements from a well established trope, whereas in Intertextuality we leave the interpretation of the borrowed speech up to the observer. A wonderful film that spoofs closed room murder mysteries is "Murder by Death", in which the subjects are all famous mystery writers who are challenged to solve a murder. Take a look at the following clip:

Movies like Murder by Death and Clue are hilarious strings of quotable forms of Intertextuality. Each line accesses imagery directly related to murder mysteries and the typical characters we find in them. Murder by Death is an ironic work since all it does is continually reference itself in a humorous way instead of trying to drive to a reveal or conclusion.

It is tempting to tease apart in detail what works and what doesn't, to find that precise inventory of mechanisms we could use in a database to drive an interactive version of a Closed Room Mystery. Unfortunately there is much left to do and many additional elements are needed that are not covered by Discourse Analysis. All this technique can do is build a global inventory of high-level means and modes we can use to drive our investigation. A number of building tasks have not been addressed here, for example Practices and Politics. Arguably these are just as important and warrant a careful examination, but those will have to wait for a follow-up post in which more concrete aspects are discussed on how a Closed-Room Murder Mystery could be created using modern video game technologies.


I hope it is clear from the preceding that the analytical techniques described above aren't meant to help you figure out the puzzle. That tends to come down to logical reasoning, and contrary to Hercule Poirot's motto that it is the psychology that matters, the outcome is usually a combination of 'this happened and here's why'. The 'because this character is such and such' is unfortunately usually left out of the conclusion, even by Agatha Christie herself. Instead the exercise of dissecting Closed Room Mysteries using Discourse Analysis can give us a better way of setting up experimental interactive narrative models where we know what we can expect from those who participate. Think of it this way: if we walk into a restaurant we expect someone task ask if we want to be seated. In terms of narrative interaction expectations we can say: if a computer game or other such simulation of a Closed Room Mystery allows a human participate to engage as one of the characters, then we have an inventory of the kinds of behaviors and thought patterns that subject might have. Even thought more can be said and reasoned about how characters will behave and be predictable, we don't have an operational model or a test environment that will allow us to see if the believes about Discourse Analysis for the types of narratives presented here holds up. For that purpose I propose a well defined and critically constrained narrative environment that could be called the Ground Hog Day Mystery. How this works will be discussed in a future post.

  1. Forster, F. (1985). Aspects of the Novel. San Diego: Harcourt Brace Jovanovich.
  2. Gee, J. (2005). An Introduction to Discourse Analysis. New York: Routledge.
  3. Van Velsen, M., Williams J., Verhulsdonck G., (2009) Narrative Concepts for AI Driven Digital Interactive Story Telling, in proceedings of the 2009 ICIDS conference
  4. Bal, M. (1997). Narratology : introduction to the theory of narrative. Toronto Buffalo: University of Toronto Press.
Suggested Reading
  1. Chatman, S. (1980). Story and discourse: narrative structure in fiction and film. Ithaca, N.Y: Cornell University Press.
  2. Genette, . Narrative discourse : an essay in method. Ithaca, N.Y: Cornell University Press, 1983.
  3. Auerbach, E. (2003). Mimesis : the representation of reality in Western literature. Princeton, N.J: Princeton University Press.
  4. Wood, J. (2008). How fiction works. New York: Picador.
  5. Bal, M. (1997). Narratology : introduction to the theory of narrative. Toronto Buffalo: University of Toronto Press.
  6. Leitch, T. (1986). What stories are : narrative theory and interpretation. University Park Pa: Pennsylvania State University Press.
* * *
* * *
On September 29th, 2013 04:49 am (UTC), (Anonymous) commented:
First of all I want to say fantastic blog! I had a quick question that I'd like to ask if you do not mind.
I was interested to know how you center yourself and clear your mind before writing.

I've had a hard time clearing my thoughts in getting my
ideas out. I truly do enjoy writing however it just seems
like the first 10 to 15 minutes tend to be lost just trying to figure out how to begin.
Any ideas or tips? Kudos!
[User Picture]
On September 29th, 2013 11:18 am (UTC), ironichles replied:
Well in a way I cheat :) By the time I sit down to write a blog post the article itself is already done in my head. For articles like this one it takes a few days or even weeks to figure it all out and by the time all the kinks have been worked out sitting down and writing it out isn't so difficult. Sometimes it helps to starts backwards and write the end first, especially if it is an essay that has a specific point or conclusion. If you're not in a hurry you can even start with nonsense and then slowly rewrite and transform it into a post. Writing in that way is kind of like the organic version of using an outline, which I never do but should.

Hope that helps.
* * *

Previous Entry · Leave a comment · Share · Next Entry