Unreliable Narration in Cinema.
Facing the Cognitive Challenge Arising from Literary Studies
University of Southern Denmark
Ever since the heyday of structuralism there has been a tendency in film studies to consider the influence of literary theory as problematic. Film is a very different semiotic system from verbal language, the sceptics say, and narratological concepts developed in the study of literary texts can only be transferred to the study of audio-visual narratives if suitable precautions are taken. And there can be no doubt that the importing of terms like 'voice', 'narrator' and 'narration' to analyse and characterise the 'telling' (yet another problematic term) of a camera has most certainly given caused problems to less theoretically minded students of film. Perhaps the differences in media are one of the reasons why a concept like 'the unreliable narrator' has been used somewhat imprecisely in film studies. Nonetheless, it seems reasonable to claim that film studies could benefit from the long and ongoing discussion of the concept that has taken place in literary studies in recent years.
As is probably well-known, there have been two major concerns here of recently: the one being that of rhetorical criticism with a focus on the ethical positioning of the reader. James Phelan is a leading figure in this connection (Phelan 2005). The other that of cognitive studies with an interest in the reader's constructive involvement in determining the unreliability which (at least until 2005 (Nünning 2005)) have had Ansgar Nünning as the prime mover.
My interest in this article is primarily what we might call 'the cognitivist challenge': One of the major insights gained from this discussion is the fact that the determination of a narrator's (un)reliability can depend to a certain extent and in some distinct cases on the conceptual framework the reader brings along to the text. In one of his most provocative articles on the issue, Nünning writes that he "will contend that we can define unreliable narration neither as a structural nor as a semantic aspect of the textbase alone, but only by taking into account the conceptual frameworks that readers bring for the text". (Nünning 1999: 60) And furthermore: "Determining whether a narrator is unreliable is not just an innocent descriptive statement but a subjectively tinged value-judgment or projection governed by the normative presuppositions and moral convictions of the critic, which as a rule remain unacknowledged." (Nünning 1999: 60) Whether or not one approves Ansgar Nünning's very generalised approach, one cannot neglect its challenges to our standard understanding of unreliable narrators with its redirection of the study towards the reader's active influence.
First of all it opens up for historical studies. The fact is that one can point to a number of examples where the detection of a narrator's unreliability or at least the extent of it seems to be governed by some historical delay. Goldschmidt's The Vicar of Wakefield is an example, as demonstrated by Vera Nünning (Nünning 1998), and one finds comparable studies of other examples carried out by Cohn (Cohn 2000) and Zerweck (Zerweck 2001). Secondly: Even though Ansgar Nünning is right in claiming that we in general are not aware of own presuppositions and moral convictions, this does not mean they have to stay unacknowledged. A part of our task as professional readers is to reflect on our readings and interpretations - i.e. to consider the frameworkwithin which they are established and perhaps reframe them - and to understand the ethical issues involved, which of course suggests an obvious bridge to the rhetorical approach to unreliable narration. And thirdly: To reject the accusations of relativism and general constructivism, focus will also have to be on the textual mechanisms and narrative conventions serving as the basis for this flexibility in the narrator's reliability. We might not be able to make an exhaustive study of this, since conventions change over time, but we can undertake partial examinations of techniques promoting ambiguity and thereby qualify the discussion further.
How these issues are to be approached in relation to cinematic narratives is, of course, a somewhat large and complicated task, so after a short overview of conception in film and literature, I will have to restrict myself to two related angles on the topic. After a brief survey of how unreliable narration has been understood in filmstudies, I will first argue how changes in cinematic narrative technique make possible the change in perception concerning narrational reliability. Secondly, I will confront the issue of how changes in ideology and historical context might influence our interpretation of a narrator. In the concluding section, I will relate the insights my brief investigation has promoted to a taxonomy of different forms of unreliability I myself have suggested and examined in relation to literature. (Hansen 2003; 2005; 2007; 2008)
Unreliable narration in film and literature
As discussed by Volker Ferenz (Ferenz 2005) in an article on the status of the concept of the unreliable narrator in film studies, the present scope has been wide and highly diverse. Seymour Chatman - one of the few who deals with film and literature equally well - uses it to describe a character who appears to be our source of the shown (i.e. in control of the image) and who turns out to be unreliable (i.e. the picture has not been true), and to describe voice-over narrators whose telling is undermined by the image-track. (Chatman 1978: 235ff, Chatman 1990: 131 ff.) These two uses are pretty much in agreement with what literary studies have been doing. But others, like David Bordwell, George M. Wilson and Gregory Currie, have applied the concept to films with non-personalised narrators where important omissions of information lead the spectator to draw his or her own or false conclusions as the film progresses (Bordwell 1985; Wilson 1986; Currie 1997), and yet others have applied it to films where the normal causal-logic of reality is suspended - either in favour of metafictional manoeuvres, as seen in Alain Resnais' L'année dernière à Marienbad (1961), or as in ghost stories like Alejandro Amenábar's The Others (2001) or M. Night Shyamalan's The Sixth Sense (1999), both partly constructed around the axis of what Tzvetan Torodov labelled the fantastic-marvellous (Todorov 1980). Ferenz shows that Tamar Yacobi's five strategies for the naturalization of textual ambiguities and inconsistencies (Yacobi 1981) is an excellent tool for sorting out some of these (mis)understandings. In Yacobi's taxonomy it is the 'perspectival principle', by which the reader brings discordant elements into a pattern by attributing them to the peculiarities of the speaker or observer through whom the world is mediated, that are congruent with what literary studies label as an unreliable narrator, and Ferenz shows that many of the films described as unreliably narrated are better understood in accordance with Yacobi's other principles - i.e. as a matter of generic or functional principles.
Thereby also hinted that in spite of considerable disagreements in literary studies concerning the theoretical framework for the concept of the unreliable narrator, no one seems to disagree on the basic issue: In literature an unreliable narrator is a intradiegetic narrator whose account of the events functions contradiegetically, i.e. in conflict with the facts or the norms and values of the fictional world (the diegesis), of which the same narrator on the other hand is our main if not only source. This conflict is manifested in different kinds of distortion in the narration, prompting the reader to read with reservations and take different constructive and reconstructive strategies with a double interest: On the one hand, to figure out the 'real' story, beyond the narrator's telling; on the other to understand the character of the narrator - her/his motivations, dispositions, psychology, ideology, etc. Even though Fludernik, in her discussion of reflectorization and figuralization in Towards a Natural Narratology, is right in claiming that "[o]nly first-person narrators can be properly unreliable" (Fludernik 1996: 213), unreliability is not restricted to first-person narrators, but is also a feature found in third-person narration. Henry James excelled in this, e.g. in the classical examples of "The Liar" and the "The Aspern Papers" (both 1888) as commented on by Booth in The Rhetoric of Fiction (Booth 1991 [1961/83]: 347-64). One might claim that cases like these are more adequately described - in the modern tongue of narratology - as 'unreliable focalization', but this of course depends on how one understands third-person narration: whether it is characterized by 'someone' (the extradiegetic narrator) telling about the incidents and characters, or whether it is the third person per se who narrates through the means of free indirect discourse, covert narration, etc. Many third-person narrations have restricted the focalization to one character, and insofar as the extradiegetic narrator does not show any superior knowledge (signs of omniscience) or delegates the perspective to other characters, it makes good sense to consider this mode as 'covert first-person narration' or 'third person proper' insofar as the extradiegetic narrator does not have any or only little significant function. In cases like these, we might very well observe unreliable reporting or judgment of the narrated events.
Subjective, objective and pseudo-objective narration
This basic feature of unreliable narrators (that is the narrator being a first-person narrator or at least internally focalized) is generally considered difficult to establish in filmic narration. Voice-over narration by a character is one solution to the problem; another the use of point of view-shots (by way of shot-reverse-shots, eyeline-matches, and subjective camerawork); a third the use of extradiegetic sound effects and music to illustrate the mental state of the focalizer; and a fourth - perhaps the most radical solution - the staging of the scenes in agreement with not only the focalizer's point of view, but also with her/his reception and understanding or - as it would be in the case of unreliability - misreception and misunderstanding.
Of these four strategies it is the voice-over and the subjective staging which most explicitly parallel the literary first-person narrator, while the use of point-of-view shots and extradiegetic sound effects and music work in a less manifest way, but nonetheless are the most common features. Furthermore, the fourth solution is in general only partially implemented, i.e. in the frame of an 'objectively told' narration, and with reference to Gérard Genette, David A. Black labels this as the use of 'pseudo-diegetic narrators' insofar as a second-degree narrative is raised to the level of the primary. (Black 1986: 22) This can be effectuated in different ways: The story-within-the-story can be set as an orally presented telling in the diegesis that is given pictorial support for the sake of the spectator. This, of course, is a very obvious means to establish and support an unreliable narrator, as seen in the extended 'lying flashback' in Hitchcock's Stage Fright (1950) or in Verbal's statement in Bryan Singer's The Usual Suspects (1995). In other cases, the character can be given the function of a voice-over narrator who comments on the depicted events and who also controls the image insofar as what is shown is in accordance with the voice-over's telling as e.g. in Nolan's Memento (2000) and Fincher's Fight Club (1999). But it is obvious that these latter two cases aren't exactly pseudo-diegetic in Genette/Black's sense, insofar as there is no primary diegetic level beyond the one accompanying the voice-over.
It is, as is well-known, an important effect of these films that during part of the narrative's progress we are 'tricked' into believing that what we are witnessing is being impersonally, zero-focalized or (as I prefer to say in what follows) 'objectively' told. This surprise effect is obtained by letting the camerawork and editing be in accordance with what we, with a potential contradiction in terms, might label 'objective perspective', even though the perspective is bound to a character: zooming and traveling cameras, where we approach the characters from a distance; establishing shots in which the characters are not present, etc.
The codes for establishing objective perspective contrast with a set of codes establishing subjective perspective. In general these latter codes are applied with a 'local function' to show that the zero focalisation of the classical cinematic invisible observer changes momentarily to a fixed focalization. We traditionally view the objective perspective as being the primary level and the subjective perspective the secondary, although this convention has often been compromised, so that the point-of-view shots are the primary level, subordinating the objective shots: stylistically they are characterised by editing and camerawork related to the objective mode of cinematic telling, but we are not shown anything that cannot be related to the domain of the focalizer.
This strategy can be implemented more or less explicitly and will of course only be significant if the focalization results in restrictions or distortions concerning what has been shown to us. An example is Ron Howard's A Beautiful Mind (2001), which tells the story of the Nobel Prize-winning mathematician John Forbes Nash, who developed schizophrenia. In the first half of the film there are a few, but nonetheless very clear, indications as to the point of view being bound to the protagonist. There are, for example, added graphics to the pictures which illustrate the protagonist's ability to detect and project patterns in-between the stars, etc., but in general we take the mode of narration as being that of the classical 'invisible observer' due to the traveling cameras often approaching Nash from a distance and having secondary characters in focus in the opening of the scenes. But in the second part of the film a large part of what we hitherto have considered part of the objective perspective (persons, actions, places) are exposed as being mental constructions and projections made by the protagonist. By shifts in the point of view, so that we now experience the story world as the other characters do, we are made aware of the fact that the major part of the film is not being objectively told, but is bound to the perspective of the protagonist. We have not only seen the events from his perspective, but we have seen what he thinks happens.
One might think that A Beautiful Mind falls into the group of films described by David H. Richter (Richter 2005) where the plot development and surprise effect depend on violations of the norms of voice and narrative coherence we usually expect films to follow (one of the best examples is Greg Hoblit's Fallen (1998) - see Richter 14f.). But a second viewing makes it clear to us that the only 'cheating' A Beautiful Mind is guilty of is letting Nash's subjective perspective be conducted in cinematic codes we at first hand see as objective, but at a second viewing understand are subjective. To exemplify this, we can take a closer look at the opening scene:
Figure 1: Stills from the opening scene of A Beautiful Mind. The repeated shot-reverse shot between the two zooming cameras disturbs the spectator's recognition of the scene being told not only from Nash's point of view but in accordance with his understanding of the events.
In the first shot the camera slowly zooms in on a professor talking to a group of students. This is followed by a reverse shot zooming out from the perspective of the professor onto the crowd, having as the focus point a person sitting at the back of the room, looking downwards. This cross-cutting is repeated, bringing us closer to respectively the professor and the person at the back (Nash), and is followed by a shot-reverse shot, where Nash looks up at one of his fellow students, who looks back at Nash (see figure 1 above).
There are several reasons for the spectator perceiving the scene as being narrated in the 'objective' mode. The repeated shot-reverse shot between the two zooming cameras disturbs the spectator's recognition of the scene being told not only from Nash's point of view but in accordance with his understanding of the events. The zooming camera does not at first hand seem bound to a character's perspective, and even though the cross-cutting follows a shot-reverse shot structure, the fact that neither the professor speaking nor Nash listening have their visual attention focused on each other makes us 'overlook' this possibility. We rely on the convention that shot-reverse shot is based on visual attention and not (as it seems to be the case here) auditive. This misunderstanding is enforced by the explicit case of subjective perspective in the final shot-reverse shot, where Nash looks up at the fellow student, who returns the look, insofar as this classical technique for promoting subjective perspective in most cases will be placed within the frame of an objective perspective. This is also how we first understand this scene. But at a second viewing, now that we are aware the perspective is Nash's, we understand the camera-work as not cheating us but as telling in a mode comparable to third-person narration with fixed focalization. And again, this is not a case of pseudo-diegetic narration insofar as the events are presented as Nash experiences them, simultaneously and depicted in accordance with Nash's reception and understanding of them.
That it is his understanding and that this understanding is not reliable is reinforced by the few shots where the point of view shifts from Nash's to other characters' - the psychologist, Nash's wife - guided by the classical signals for point-of-view shots as seen in the sequence of shots in figure 2 and 3. In figure 2 the first shot establishes Nash's point of view. He first sees his wife Alicia, then the imaginary Parcher. When Alicia hears her husband talk to someone in the room, she looks in the direction he is looking. The camera moves so that we can see what Alicia sees - which is nothing - before returning to Nash's perspective. In figure 3 we first see Nash observe the imaginary Marcee. Then the perspective is reversed, and even though Nash is the only one looking in the direction of Marcee (now invisible), it is clear to the spectator that it is the perspective of the two other present characters, the psychologist Dr. Rose and the wife Alicia, we are being given - basically cued by Dr. Rosen's brief look over his shoulder and his question to Nash of whether he sees the imaginary persons at the moment.
Figure 2: Alicia's point of view framed by Nash's.
Figure 3: John Nash is watching the imaginary Marcee, while she does not exist in the perspective of the two other present characters, the psychologist Dr. Rose and Nash's wife Alicia.
Though very common, this stabilization of the unreliability of the narration is by no means general. Roman Polanski's Rosemary's Baby (1968) is an example on how the spectator at first hand is presented with what seems to be objective narration, but gradually becomes aware that what we are witnessing might be Rosemary's fantasies. As in the case of A Beautiful Mind, we are given some clearly marked subjective shots and scenes (we witness one of her dreams and the rape scene is filmed and cut from Rosemary's perspective), but the main part of the film is narrated in the conventional mode of objective telling, without, however, leaving Rosemary's point of view or 'domain', as we could say - to reject the visual aspect of this metaphor. The spectator is never given a final clue to determine whether Rosemary is right or wrong in her belief that she is having the devil's child. Instead, the story is ambiguous and left open to the spectator's interpretation.
We might label the strategy of letting a subjective perspective narrate in the codes of objective narration a 'pseudo-objective narrator'. Furthermore, we can gain a better understanding of what is at stake in this mode of telling if we compare with the literary use of first-person present tense (FPPT), i.e. what Cohn labels 'simultaneous narration'. Cohn draws attention to three central and interdependent characteristics of simultaneous narration in literary texts. Firstly, the narrative situation is incongruous insofar as it cannot be naturalised by the reader's attempt to postulate a verisimilar narrative situation (tape-recording, oral diary, etc.). We can draw a parallel between this and the seemingly indeterminacy between a subjective and objective mode of telling that we find in pseudo-objective narration. Secondly, the semantic implications of the narrative tense, e. g. the fact that the narrator's imaginings and observations are being juxtaposed and presented in the same temporal grammar, facilitate a high degree of uncertainty regarding what is true and false, real and fantasy within the fictional world. This is comparable with the case of Rosemary's baby. And thirdly, "the absolute focalization of its narrated experience" rests on the analogy between action, thought and narration. To Cohn's list we can add a fourth characteristic, namely that FPPT promotes a complete elimination of any visible signs of an authorial or narratorial agent 'beyond' the first-person narrator.
As noticed by Fludernik, the extended use of first-person present tense is a relatively new occurrence in literary fiction (Fludernik 1996: 223), and it has an interesting consequence for the reader's "(re-)conceptualization of the natural storytelling frame, where a story has to have happened in the past in order to become tellable". We can make a similar claim regarding the pseudo-objective narrator in cinematic narration, since the spectator has to reconceptualise the conventional cinematic storytelling frame, where subjective telling normally is codified explicitly. In pseudo-objective narration the subjective point of view is not necessarily marked explicitly, and considered in the light of the increased use in contemporary cinema of this form we might very well observe a change in our receptive schemes, so that our primary response to the conventional codes of objective telling is to consider them as a potential subjective perspective. And thereby open up for a reflection on the potential reliability of the narrator.
So much for the first aspect I wanted to touch upon. The second concerns the change in ideology and historical context, and its potential effect on our understanding of a narrator's reliability.
I will here turn my attention towards a film which at first seems as if it is being told in an conventional objective mode, but by the very end opens for another possibility.
Florian Henckel von Donnersmarck's Das Leben der Anderen (2006) is concerned with the period of the cold war and divided Germany. The story takes place in East Berlin in 1984, where the Stasi agent Wiesler, a convinced supporter of the communist regime, is assigned to monitor playwright Georg Dreyman, who is suspected of Western leanings. Secretly, Dreyman's apartment is being bugged, and Wiesler himself is in charge of the surveillance. But Wiesler soon finds out that the real reason why Dreyman is being spied on is that a minister is attracted to Dreyman's girlfriend, actress Christa-Maria; if Dreyman is arrested, the minister will have free rein. This demotivates Wiesler.
Dreyman is a supporter of the regime, but dislikes the way dissidents are treated. When an artist friend commits suicide because of his marginalisation, Dreyman anonymously publishes an article in West Germany on the neglect of suicide rates in the GDR.
Wiesler observes this development, but via his monitoring has gained an understanding of Dreyman, and starts lying in his reports in order to protect him. When Christa-Maria is arrested for drug abuse, she turns Dreyman in, but Wiesler removes the evidence from the apartment before the search team arrives. Unaware of this, Christa-Maria walks away in shame, and dies when she is hit by truck. As a result, the operation becomes pointless; hence Wiesler's chief officer, Anton Grubitz, calls it off and ensures the end of Wiesler's career.
At the end of the film, after German reunification, Dreyman finds out the truth while searching through his file in the Stasi archives. He finds out Wiesler's location and sees that he has meanwhile become a distributor of leaflets. Dreyman does not approach Wiesler, but writes a book called 'Die Sonate vom Guten Menschen' dedicated to "HGW XX/7" (Wiesler's Stasi code name). When Wiesler buys the book and the bookseller asks him if he is to wrap it as a present, Wiesler responds: "Nein. Es ist für mich."
Even though critical voices were raised against the tendency to caricature, Das Leben der Anderen has been highly praised for its realistic depiction of the horrors of the former communist state, and this is of course a very important layer in the film. But one cannot but be astonished by the fact that no critic has paid attention to the complexities and ambiguities raised in the final sequence about Dreyman writing a novel about the incidents since an obvious conclusion for the spectator is that what we have witnessed, the film we have seen, is this novel, so to say, that is: Dreyman's recollection of what happened then and his reconstruction of the events. And when one follows this strategy, the film turns into a rather different story.
First of all the status of Dreyman's novel is rather dubious: Dreyman makes an effort to find Wiesler, but he never approaches him to check whether his understanding of the events is right. Spectators who trust the text would probably claim that this is due to the fact that Wiesler at one and the same time was Dreyman's persecutor and saviour, and that Dreyman's gratitude to Wiesler doesn't include forgiveness. But considered from the perspective of the film being told by Dreyman, there are more obvious reasons - namely that contacting Wiesler on this issue might disturb the picture he himself has constructed. The fact is that Dreyman must have written his book based on his own memory and what he learned from studying his file in the Stasi archive. Here, however, there is no evidence of e.g. Maria being forced to turn him in or of Dreyman turning from a regime-supporting artist into an undercover dissident, or of Wiesler's motivation for covering Dreyman - if he actually did cover anything. All these events are - if the film is a depiction of Dreyman's novel - pure speculation and construction.
This of course doesn't change the general theme of the film - i.e. what the effects of suppression and totalitarian systems are. But it adds another layer to this story, one which we might approach by way of Phelan's ethical criticism.
According to Phelan, ethical criticism is a matter of binding "ethical response to the techniques of narrative itself", by focusing "on the links among technique (the signals offered by the text) and the reader's cognitive understanding, emotional response, and ethical positioning." (Phelan 2005: 22) Phelan goes on to specify this ethical positioning as being established through the dynamic interaction between four ethical situations, that of the characters, that of the narrator, that of the Implied Author, and that of the empirical reader. I am not going to get involved in the never-ending discussion of whether it makes sense to operate with an Implied Author in film - but instead intend to make a swift adjustment to Phelan's conception and substitute 'Implied Author' by a term suggested by Manfred Jahn: 'FCD' - the Filmic Composition Device being the corporate principle for filmic communication.
If Das Leben der Anderen is considered as objectively told, one would say that the ethical positioning primarily takes place between the reader, the characters and the FCD, insofar as the narrator is anonymous and the characters function as rather flat and stereotyped figures, with the exception of the two protagonists developing from bad guy into good guy (Wiesler), and from not all that good a guy to good guy (Dreyman). Everyone else plays their more or less expected role - that is of puppets in the hands of FCD in a piece about the value of sincerity, goodness and humanity.
If the movie is told by Dreyman, the ethical positioning changes from the FCD level to that of the narrator (being Dreyman) and his relation to the person he was earlier, according to the Stasi files (the regime-supporting artist), and the one he want us to see he was (a good human being), which also might be indicated by his name: Dreyman, drei man, three persons; and at the same time the active interaction of the reader is being stressed. We are put in a position of dynamic interpretative response, where we not only are supposed to perceive a story being told but have to reconstruct it actively by negotiating and evaluating what we have been told. In that sense we are not only recipients of a story, but also re-tellers insofar as we, as soon as we realise that what we have been watching might not be the 'real' story, perform an act of re-evaluation of the story. When we do so, we find a lot of details make better sense than they did in our former approach: e.g. that Stasi and the minister are caricatured (we laugh at their stupidity in spite of the seriousness of the topic), and that the only person who seems to be a sincerely good man (vis-à-vis the title of Dreyman's book) is Dreyman himself: He shows understanding towards the relationship between his fiancée and the minister; he understands the dissidents, but balances himself somewhere in-between loyalty and responsibility. We are also given a reason for him not contacting Wiesler: If he made contact with him, he would no longer be in absolute control of the story. From this perspective another dimension is added to the title of the movie: Das Leben der Anderen not only refers to the border between dissidents and the regime supporters - but also refers to the life Dreyman would liked to have lived - but that he failed to accomplish.
My aim with this - rather schematic - reading of Das Leben der Anderen is not to claim that the first suggested understanding of the film is wrong. Rather the point is to show that the film is open to both readings because of a potential instability in the narration. The question of why no critic of the film has detected this potential and, would I say, plausible understanding of the film is impossible to give a clear answer to. But we might at least achieve some understanding by comparing with similar texts where the trust in the narrator's account has changed utterly - like for instance Marlow's from Conrad's Heart of Darkness. As most of you know, it was not until 1958, with Guerard's book Conrad the Novelist, that we began to realise "that the story is not primarily about Kurtz or about the brutality of Belgian officials but about Marlow its narrator" (Guerard 1958: 37). - Perhaps because the brutality of imperialism was very urgent when the first readings of the story was made, just as the horrors of the socialist regime are in the contemporary context of Das Leben der Anderen. As time passes, the grey shades of initially black-and-white historical tragedies become more visible to us; - and perhaps even more urgent for us to recognise.
My aim with this paper has not been to claim a general interpretative relativism, but to open up for an active use of the insight gained by the cognitive turn in the study of unreliable narration. While/Whereas the initial discussion between 'rhetoricians' (Phelan) and 'cognitivists' (Nünning) of what determined narration as reliable or unreliable was formulated as an 'either-or' (either it was the Implied Author or it was the reader), we are now moving in a more co-operative direction, considering it a 'both-and' issue. Unreliable narration can be part of an intentional strategy used by the author, but it can also be a result of the reader's understanding of the narrator. We might, of course, follow e.g. Cohn and suggest a conceptual distinction between 'unreliable' and 'discordant' narration, saving the latter for the cases where it most evidently is a matter of ideological changes that causes our experience of a narrator's unreliability. But the question is whether we actually can keep these two categories apart? Firstly, we have to take into account the effect is the same in both cases. Secondly, as I have attempted to show in the first part of this presentation, narrative techniques are not universals but can be used for different purposes and change through time.
So what we have to acknowledge is that the effect of unreliable narration has many sources. I have elsewhere suggested a taxonomy of four categories - intranarrational, internarrational, intertextual and extratextual unreliability - which I will briefly include here with reference to how they are enacted in cinematic narrative:
Intranarrational unreliability refers to unreliability established and supported by the large stock of what Wall has labelled "verbal tics" and other textual indicators of uncertainty in and on what is being narrated: explicit claims of reliability, subjective efforts at partiality, absurdities or violations of the logic and the premises of the story world, changes in the use of pronouns, a marked attention towards one's reactions, etc. In filmic narrative these 'tics' are comparable to the cases where the staging is evidently expressing a disturbed, subjective point of view as in e.g. Cronenberg's Spider. But as both Marie-Laure Ryan (Ryan 1981) and Wall have noted, this conception is not without problems since, the discursive indicators which are directing our attention towards the reliability of the narrator, do not necessarily make the narrator unreliable - and we might therefore have to differentiate between the unreliable narrator and unreliable narration. The two instances would obviously in most cases accompany each other - but not necessarily. This can briefly be demonstrated by bringing in Cameron Crow's Vanilla Sky (2001): In Vanilla Sky the focalizer and protagonist seemingly develop schizophrenic and psychotic traits after a car-accident. He, for example, seems to confuse the woman he loves with the woman who was involved in and died due to the car accident- cinematically illustrated by letting the woman change her appearance when he is together with her. To the spectator, who observes these shifts within the frame of the accident, the trauma which followed it and the general instability with which the protagonist is characterised, these changes in appearance are naturalised by being considered as illustrations of his still aggravated mental condition. But at the end of the movie, another and (seen from the logic we handle reality by) much more incredible explanation is given: The protagonist has accepted a company's offer to be frozen and live on in a computer-controlled mental space - an asylum with the one he loves and without the physical failings the accident gave him. His psychotic experiences are provoked by an error in the computer software - and can be corrected at his request. To conclude in relation to the topic of this presentation: Even though his discourse carried all the indicators we would connect with the unreliable narrator, it was only his narration that was unreliable - he, as an existent in a story, was not. In A Beautiful Mind we find the opposite situation, being where the focalizer's discourse doesn't carry any direct indications of the unreliability, but instead constructs a totally coherent and at first glance reliable account, but which during the discourse is contradicted and disclaimed by other narrational instances. In contrast to Vanilla Sky this movie is about a person who actually develops schizophrenia. But in the first, long period of his sickness, there is nothing which reveals it to the spectator. In relation to the present discussion, the interesting point is that the discourse doesn't mark the unreliability at first, and not until later disclaims it. - In other words: No discursively marked unreliability (no unreliable narration), but nevertheless an unreliable focalizer.
With Internarrational unreliability I refer to the situation where a narrator's version of the incidents is contrasted by the version of one or more other narrators. In filmic narration we find this strategy used in cases of voice-over narration where the narrator's comments contrast with what the pictures show us, and in cases with distributed point-of-view shots, as it was the case of A Beautiful Mind. Also Kurosawa's Rashomon (1950) belongs to this category: The film depicts the rape of a woman and the apparent murder of her husband through the widely differing accounts of four witnesses, including the rapist and, through a medium, the dead man. The stories are mutually contradictory but supported by a pictorial evidence of what the witnesses tell us, leaving the viewer to determine which, if any, is the true account.
Intra- and internarrational unreliability are alike in that they are both based on textually observable issues, which are manifested as conflict. When it comes to the two remaining categories, they differ from the intratextual variants since they depend on extra- or contextual issues. Intertextual unreliability is based on manifest character types, which direct the reader's attention towards their potential unreliability. Rosemary's Baby is a good example of this strategy insofar as the whole determination of Rosemary as being unreliable rests on our assumption that she is victim of some kind of pregnancy psychosis. As discussed earlier, Rosemary's Baby is told in the pseudo-objective narrative mode, and the narration therefore does not have any intratextual markers of unreliability - omitting/except for the satanic rituals at the end, of course. But considered as a horror-movie these scenes are totally reliably told. So, if we consider the film an example of unreliable narration, it is partly due to the character-type Rosemary is, since the mere identification of a character-narrator as a specific, recognizable type that normally would act unreliably will let the reader pursue this aspect in the reading.
My final category, the extratextual variant, designates unreliability that depends on the reader's direct implementation of his or her own values or knowledge of the story world. This was what we practised in Das Leben der Anderen when we considered how the film, which at the time of its appearance was considered to be straightforwardly reliable, can read as unreliable in a new historical situation (due to changes in ideology and culture).
As I have argued elsewhere, the four basic forms of narrational unreliability I have outlined here will of course often function together in a text, so that the unreliability is simultaneously marked in the narrator's discourse or focalizer's perspective, by other narrators' or focalizers' telling or perspective, by virtue of the character type the narrator or focalizer is modelled on, and in relation to the knowledge the reader brings to the text. It is thus possible to determine four different strategies, but they should - because of their very divergence - be treated as distinct features promoting the same effect, and their historical and cultural variability should be studied.
Black, D. A. (1986). >Genette and film. Narrative level in the fiction cinema.< Wide Angle 8(3/4): 19-26.
Booth, W. C. (1991 [1961/83]). The Rhetoric of Fiction. London et. al., Penguin Books.
Bordwell, D. (1985). Narration in the fiction film. Madison, Wis., University of Wisconsin Press.
Chatman, S. (1978). Story and Discourse. Narrative Structure in Fiction and Film. Ithaca & London, Cornell University Press.
Chatman, S. (1990). Coming to Terms. The Rhetoric of Narrative in Fiction and Film. Ithaca & London, Cornell University Press.
Cohn, D. (1999). The Distinction of Fiction. Baltimore & London, The Johns Hopkins University Press.
Cohn, D. (2000). >Discordant Narration.< Style 34(2): 307-316.
Currie, G. (1997). Image and mind film, philosophy and cognitive science. Cambridge, Cambridge University Press.
Ferenz, V. (2005). >Fight clubs, american psychos and mementos. The scope of unreliable narration in film.< New Review of Film and Television Studies 3(2): 133-159.
Fludernik, M. (1996). Towards a 'Natural' Narratology. London; New York, Routledge.
Hansen, P. K. (2003). >Hvem sagde upålidelig? Om sanktioneringen af upålideligheden i den upålidelige narration<. Berättaren: En gäckande röst i texten. L.-Å. Skalin. Örebro, Örebro Universitet: pp.149-176.
Hansen, P. K. (2005). >When Facts Become Fiction: On Extra-Textual Unreliable Narration<. Fact and Fiction in Narrative: An Interdisciplinary Approach. L.-Å. Skalin. Örebro, Örebro University: 283-307.
Hansen, P. K. (2007). >Reconsidering the Unreliable Narrator.< Semiotica(165 (1/4)): 227-246.
Hansen, P. K. (2008). >First Person, Present Tense. Authorial Presence and Unreliable Narration in Simultaneous Narration<. Narrative Unreliability in the Twentieth-Century First-Person Novel. E. D'Hoker and G. Martens. Berlin, De Gruyter: 317-338.
Liptay, F. and Y. Wolf (2005). Was stimmt denn jetzt? Unzuverlässiges Erzählen in Literatur und Film. München, Ed. Text und Kritik.
Nünning, A. (2005). >Reconceptualizing Unreliable Narration<. A Companion to Narrative Theory. J. Phelan and P. J. Rabinowits. Malden, Ma & Oxford, Blackwell Pub.: 89-107.
Nünning, V. (1998). >Unreliable narration und die historische Variabilität von Werten und Normen: The Vicar of Wakefield als Testfall für eine kulturgeschichtliche Erzählforschung<. Unreliable Narration. Studien zur Theori und Praxis unglaubwürdigen Erzählens in der englischsprachigen Erzählliteratur. A. Nünning. Trier, Wissenschaftlicher Verlag Trier: 257-285.
Phelan, J. (2005). Living to tell about it. A rhetoric and ethics of character narration. Ithaca, Cornell University Press.
Richter, D. H. (2005). >Your Cheatin' Art: Double Dealing in Cinematic Narrative.< Narrative 13(1): 11-28.
Ryan, M.-L. (1981). >The Pragmatics of Personal and Impersonal Fiction.< Poetics(10).
Thomson-Jones, K. (2007). >The literary origins of the cinematic narrator.< British Journal of Aesthetics 47(1): 76-94.
Todorov, T. (1980). The fantastic. A structural approach to a literary genre. Ithaca, N.Y., Cornell University Press.
Wall, K. (1994). >The Remains of the Day And Its Challenges to Theories of Unreliable Narration.< The Journal of Narrative Technique 24(1): 18-42.
Wilson, G. M. (1986). Narration in light. Studies in cinematic point of view. Baltimore, Johns Hopkins University Press.
Yacobi, T. (1981). >Fictional Reliability as a Communicative Problem.< Poetics Today 2(2): 113-126.
Zerweck, B. (2001). >Historicing Unreliable Narration: Unreliability and Cultural Discourse in Narrative Fiction.< Style 35(1): 151-178.
 For an excellent account of the problems connected with the concept of 'the narrator' in film studies, see Thomson-Jones 2007.
 This very broad use of the term is seen in the collection made by Liptay and Wolf 2005
 Cohn 1999 - see also my own article on the issue Hansen 2008.
 Among the films that use this form of telling we find David Cronenberg's Spider (2002), Joe Wright's Atonement (2007), Brad Anderson's The Machinist (2004), Alejandro Amenábar's The Others (2001), David Fincher's Fight Club (1999), M. Night Shyamalan's The Sixth Sense (1999), a.o.
 One might object that this transition doesn't find motivation due to the media difference. But it is a strong convention in cinema that spoken and written material is represented in film with the aid of the visuals. Hitchcock's Stage Fright is one example; Jo Wright's film version of Ian McEwan's Atonement (2007) another.
 The categorization is sketched out and developed in Hansen 2003, Hansen 2005, Hansen 2007.
 For a systematic and detailed account, see Wall 1994 Allrath (1998). These indicators function as cues for the reader for reflection over the trustworthiness of the narrator, which, however, finally has to result in a general rejection of the narrator's version of the story in favour of a (re)construction of 'how it really was' to label the narrator unreliable.