Apply a particular aspect of Bakhtin's theorising to a short extract from a novel of your choosing and give an account of the results.

Creative Commons License
This work is licensed under a Creative Commons Attribution-Share Alike 2.5 License

Bakhtin's theorizing has been appropriated in a number of ways by different academic disciplines, and not always in an entirely accurate manner. However this process is a part of understanding, new meanings may be revealed in Bakhtin's ideas that he did not realize were there:

A meaning only reveals its depths once it has encountered and come into contact with another, foreign meaning: they engage in a kind of dialogue, which surmounts the closedness and one-sidedness of these particular meanings...[note1]

In this essay I shall attempt to appropriate the voice of Bakhtin into the new speech genre of interactive fiction. By viewing the genre of Bakhtinian thought from external position of interactive fiction it is possible that some of the,'semantic possibilities that lie within,'[note2] his writing may be revealed.

The phenomenon of the World Wide Web has brought to the attention of most of the population of the developed world the possibilities of communicating by computer. It is now possible, among other things, to order pizza, to apply for jobs, to do training or to buy groceries using a computer and a modem. All the activities of our daily lives can be carried out, if we so wish, without leaving our living room. Aside from the new commercial opportunities made available by this new medium of communication there are also new possibilities for art. Art no longer has to be a static object, we can interact with it. Interactive fiction has yet to produce any generally acknowledged literary masterpieces, it has yet to transcend it's roots in roleplaying and computer games, so I will be looking at the genre as a whole and describing how Bakhtin's ideas illuminate the possibilities available to us. There currently exist two main stylistic lines of interactive fiction, hypertext[note3] and character based interaction[note4]. I will examine these two lines and the applicability of Bakhtin's concept of polyphony.

A hypertext system is essentially a set of documents and a set of links between them, navigation of the documents is not determined in advance by the author(s) but is decided upon by the reader. A reader may decide to add his own documents or links, which may affect the future interpretation by other readers. When taken to the extreme each word in a document can be a link to another document, or several other documents, and each reader can also be an author. George Landow has commented on the applicability of Bakhtinian ideas:

In terms of hypertextuality this points to an important quality of this information medium: hypertext does not permit a tyrannical, univocal voice. Rather the voice is always that distilled from the combined experience of the momentary focus, the lexia one presently reads, and the continually forming narrative of one's reading path.[note5]

He even goes so far as to credit Bakhtin with inventing hypertext, claiming his,"description of the polyphonic literary form presents the Dostoevskian novel as a hypertextual fiction in which the individual voices take the form of lexias[a]."[note6] This is very suggestive but, in my opinion, not entirely accurate - a lexia is not equivalent to a Bakhtinian voice. Limiting a voice to a single lexia would be like allowing a character only one speech, the character would have no option of a future reply to another lexia. It appears to me that a lexia has more in common with an utterance, a voice may make many utterances in the course of a dialogue. Thus the very nature of hypertext inclines it towards dialogue, it exists as a set of dialogized utterances and the possibility of adding links means that every word means that we can make use,"of someone else's discourse for [our] own purposes by inserting a new semantic intention into a discourse which already has, and which retains, an intention of its own."[note7] Every word becomes double voiced. Every utterance can be dialogized.[note8]

We have established that hypertext has the necessary properties required for polyphony. Hypertext can be viewed as the concrete instantiation of dialogized heteroglossia except the links are there for everyone to see, the scaffolding is left in place, so to speak. However, dialogization and double voiced words alone do not make a polyphonic work of literary art. In their discussion of polyphonic novel forms G. Morson and C. Emerson argue that,"polyphony is in fact a theory of the creative process. A polyphonic work must be created in a special way..."[note9] Traditionally the creative process has been viewed as either the capturing of some divine inspiration or the methodical procession from plan to finished work. In contrast,"polyphonic creation is an open process that seeks 'surprisingness' at almost every step of the way."[note10] Dostoevsky is seen as being exemplary of this process. The starting point is voice, voices are imagined, characters in the book, and then the task of the author is to provoke them into dialogue with each other and his own narratorial voice. These dialogues may suggest new situations or refinements to the voices of particular characters. The plot is determined, in a large part, by the results of the previous dialogues rather than in advance by the author. The goal of the plot ceases to be the finalization of the hero and becomes instead the maximising of the dialogic potential of the voices. The polyphonic novel is intended,"to turn readers not into analyzers of the character, action and circumstance, but into dialogic partners of the characters and author."[note11] The author himself is not in a privileged position regarding these dialogues, he does not know in advance how they will turn out (If he did, if he was bending the dialogue of the characters towards some predetermined conclusion, then we would have a monologic text, merely the image of polyphony rather than the thing itself). There are some problems, endings are antithetical to polyphony but need to be incorporated into an actual novel. How can the reader be an equal partner in the dialogue of the novel when they cannot actually participate in the dialogue within the novel? The move to hypertext answers the latter question but the problem with endings continues, it is often the case in current hypertext works that the ending occurs when the user chooses to stop interacting rather than at any specific point. Also, the author must have,"such a vivid sense of the characters that he can truly address them."[note12] This seems to require, of the great polyphonic author, that he be somewhat schizophrenic, a high price to pay. The second stylistic line of interactive fiction is directly concerned with this question.

MUDs[b] are virtual spaces populated by up to ten thousand characters, each created by one of the players. You are allowed to move about the space, build new areas and converse with other characters. Goals vary but in typical games they are open ended (or unfinalizable) and largely determined by the personal preferences of the players. Appendix I is taken from Sherry Turkle's book,"Life on the Screen".[note13] The book is primarily concerned with psychology rather than narrative but is still useful to us. Doug illustrates the potential for polyphonic authorship when he simultaneously operates several characters in different environments, it doesn't seem a large step to operate those characters in one environment and allow those characters to take part in a dialogue with each other.

A feature of most MUDs are NPCs[c], characters that are controlled by the computer rather than a live human player. Some are there simply to provide directions or offer help when asked and could never be mistaken for a human player but some are far more complex. Appendix II is another extract from Sherry Turkle's book and describes a particular NPC called Julia. An NPC is a computer controlled character, essentially a computer program. Julia is a descendent of Eliza, a program written in the late sixties with the goal of passing the Turing Test[note14]. These programs work by recognising keywords and particular forms of sentences and having a set of preset responses and response forms. Janet Murray discusses the limitations of such programs:

It is easier to create a fictional personality than it is to model knowledge of the world. In fact, Julia is more imaginatively present when she is less intelligent, because her sense of presence does not come from giving factually correct information but from demonstrating dramatically appropriate behaviour. The MUD provides a social framework in which her formulaic responses make sense.[note15]

This is not just a problem for computer programs, as Bakhtin points out:

Many people who have an excellent command of language often feel quite helpless in certain spheres of communication precisely because they do not have a practical command of the generic forms used in the given spheres ... this is entirely a matter of ... the lack of a sufficient supply of those ideas about the whole of the utterance that help to cast one's speech quickly and naturally in certain compositional and stylistic forms ...[note16]

Murray describes various attempts by her students to build characters with which it is possible to have conversations, some of which are successful in certain proscribed domains. Their success depends on maximising the 'Eliza Effect','the human propensity to suspend disbelief in the presence of a persuasive dramatic presence.' In a sense all literature depends upon this to some extent, when Bakhtin imputes agency to Dostoevsky's characters he is making a claim about how real they seem to us. Characters in a polyphonic novel are, in one sense, more limited than the computer characters mentioned above because their responses for the reader are forever finalised, they will never say anything else. Computer programs present the possibility of unfinalisability, they will always be able to say something else, to continue the discourse. Against this one advantage over traditional literature these computer programs have two main disadvantages, they lack an inner life and they lack the possibility of character development. Some programs have attempted to rectify the first complaint by having emotional states not visible to the interlocutor[note17]and, more recently, attempts have been made to create characters with goals and agendas, though so far they've only managed to produce realistic cats and dogs.[note18] The lack of an inner life may contribute to the second problem, how can characters develop when there is nothing in them to develop. Even the most recent creations only move between a finite and fairly small number of states. For Bakhtin the possibility for development, unfinalisability of character as well as dialogue, was essential in polyphonic heros. He says:

All the stable and objective qualities of a hero - his social position, the degree to which he is sociologically or characterologically typical, his habitus, his spiritual profile and even his very physical appearance - that is, everything that usually serves an author in creating a fixed and stable image of the hero, 'who he is', becomes in Dostoevsky the object of the hero's own introspection, the subject of his self-consciousness; and the subject of the author's visualization and representation turns out to be in fact a function of his self- consciousness.[note19]

For Bakhtin the hero of the novel is never fully formed at the outset. The computer characters seem more like Bakhtin's description of heros of,'high distanced genres,' where the hero is,'an individual of the absolute past and of the distanced image. As such he is a fully finished and completed being.'[note20] For a hero to be an equal partner in dialogue it must have the same possibilities for growth, whether computers can provide these possibilities is a matter of open debate. There is a consensus that leaving computer scientists to their own devices will not alleviate the problem, character development is traditionally the domain of the literary arts so some sort of collaboration seems to be in order if we are to usher in the brave new world of interactive fiction. As Stern and Frank conclude:

We've found that developing lifelike, believable characters requires people with interdisciplinary skills from both technology and the arts. Until recently, building lifelike artificially intelligent characters has been treated as a computer science problem. Character development, in any medium, is fundamentally an artistic, aesthetic decision-making process.[note21]

Murray agrees that,'new narrative techniques can develop,' if,'software environments, created and refined by programmers working in collaboration with writers were more widespread.'[note22]

In conclusion, it is apparent that the Bakhtinian concept of polyphony and the polyphonic novel has much to offer interactive fiction. It seems to be embodied to a large extent in hypertext thanks to that genre's predisposition towards dialogic forms and double voiced words. By contrast the second style of interactive fiction lacks the obvious double voicedness but offers extended forms of unfinalisable dialogue. It seems likely that the great masterworks of interactive fiction, should they appear, will have to combine these two sub genres to a great extent, perhaps double voiced narrative will need to be somehow incorporated into the inner life of computer generated characters, but will still owe a great debt to the theories of Mikhail Bakhtin.


The original appendices for this essay were sections from Sherry Turkle's book, 'Life on the Screen', and are not reproduced here. I recommend you go and buy it!


a. Landow borrows the term 'lexia' from Barthes. [back]

b. Multi User Domains (originally Multi User Dungeons, from the usual location of most early systems). [back]

c. Non Player Characters. [back]


1. M.M. Bakhtin 'Speech Genres and Other Late Essays' trans. Vern W McGee, University of Texas Press, 1986. p7. [back]

2. ibid. p5.[back]

3. The most widely acknowledged example is 'Afternoon' by Michael Joyce. Links to this and other works can be found at ''.[back]

4. Andrew Stern and Adam Frank 'Multiple Character Interaction Between Believable Characters' available on the world wide web at:
Also see:[defunct]

5. Landow, George P. 'Hypertext: The Convergence of Contemporary Critical Theory and Technology' Johns Hopkins University Press, 1992. p 11. [back]

6. ibid. [back]

7. M M Bakhtin 'The Problems of Dostoevsky's Poetics' trans. Caryl Emerson. University of Minnesota Press, 1984. p 189. [back]

8. It is interesting to note, at this point, some of the differences between the idealized hypertext system I have described and the its real world instantiation as the world wide web. In an ideal hypertext system any user can add his own documents and his own links to already existing documents. As it stands, the world wide web allows you to add links only to your own documents. This situation seems unlikely to change, large companies seem unwilling to allow their monologic documents to be appropriated for heteroglossic use. Software which allowed users to add commentary to other documents, which could then be viewed by other users of the software, was recently released. The aforementioned large companies immediately started legal action... [back]

9. Gary Saul Morson and Caryl Emerson 'Mikhail Bakhtin : Creation of a Prosaics' Stanford University Press, 1990. p 243. [back]

10. ibid. p244. [back]

11. ibid. p249. [back]

12. ibid. p246. [back]

13. Sherry Turkle 'Life on the Screen' Phoenix, 1997. [back]

14. The Turing Test was proposed by Alan Turing in his paper 'Computing Machinery and Intelligence' (Mind LIX, no. 2236 (Oct 1950) pp433-460), as a test for machine intelligence. An independent judge has a conversation, via teletype, with a male and a female, his task is to decide which is which. Under these conditions if the male is replaced with a computer will the judge be able to spot that it is a computer? Turing argued that going undetected in this test would prove machine intelligence, a claim still hotly disputed. It is important to note that the judge is distracted from the primary purpose of the test. [back]

15. Janet H Murray 'Hamlet on the Holodeck : The Future of Narrative in Cyberspace' MIT Press, 1997. p218 [back]

16. M.M. Bakhtin 'Speech Genres and Other Late Essays' trans. Vern W McGee, University of Texas Press, 1986. p80. [back]

17. The classic example is PARRY, a computer generated paranoid schizophrenic who was able to convince several doctors (who were trying to diagnose several patients via teletype) that it was a real (paranoid) person. See Janet Murray 'Hamlet on the Holodeck' (MIT 1997) pp222-226. [back]

18. The fruits of these labours are available as the commercial programs Catz and Dogz, they are described in Stern and Frank's paper 'Multiple Character Interaction Between Believable Characters' available on the world wide web at [back]

19. Here we see an important difference between polyphony as Bakhtin described and this particular style of interactive fiction, for Bakhtin the self- consciousness of the heros are a function of the author's self-consciousness while in the interactive arena it is a function of the reader's self-consciousness. This perhaps indicates that in interactive fiction the roles of author and reader will be re-examined and changed. (Quote from M M Bakhtin 'The Problems of Dostoevsky's Poetics' trans. Caryl Emerson. University of Minnesota Press, 1984. p48.) [back]

20. M M Bakhtin 'The Dialogic Imagination : Four Essays' trans. by Caryl Emerson and Michael Holquist. University of Texas Press, 1981. p34. [back]

21. Andrew Stern and Adam Frank 'Multiple Character Interaction Between Believable Characters' available on the world wide web at: [back]

22. Janet H Murray 'Hamlet on the Holodeck : The Future of Narrative in Cyberspace' MIT Press, 1997. p222. [back]

© Robert Crowther