CC BY
7
Эпистемология и философия науки 2022. Т. 59. № 2. С. 38-42 УДК 167.7
Epistemology & Philosophy of Science 2022, vol. 59, no. 2, pp. 38-42 DOI: https://doi.org/10.5840/eps202259218
A possible-worlds construal of unreliability in film
Dorit Abusch - PhD, Professor. Cornell University. 203 Morrill Hall, Ithaca, NY, 14853, USA; e-mail: da45@cornell.edu
This paper comments on Emar Maier's "Unreliability and point of view in filmic narration". It is suggested that, without having discourse representations that include embedding operators, films can be unreliable in the broad sense of having propositional contents that depart from inferable, realistic scenarios. Second, films and embedded shots in film can convey agent-centered information without being composed of point-of-view shots. The reason is that the discourse representation can include information about discourse referents that identifies a depicted individual as a counterpart of the experiencer.
Keywords: De se, discourse representation theory, embedding, film, point of view, unreliable narrator
Возможные миры истолкования
ненадежности фильмов
Дорит Абуш - доктор философии, профессор. Корнелльский университет. 203 Morrill Hall, Итака, 14853, Нью-Йорк, США; e-mail: da45@cornell.edu
В этой статье комментируется статья Эмара Майера «Ненадежность и точка зрения в киноповествовании». Во-первых, обосновывается положение, согласно которому без дискурсивных репрезентаций, содержащих встроенные операторы, фильмы могут быть ненадежными в широком смысле, т.е. иметь пропозициональное содержание, отходящее от предполагаемых реалистичных сценариев. Во-вторых, фильмы и встроенные кадры могут передавать информацию в перспективе агента и в отсутствие кадров, воспроизводящих эту перспективу. Происходит это в силу того, что репрезентация дискурса может включать в себя информацию о дискурсивных референтах, которые предъявляют изображаемого индивида в качестве двойника носителя переживания. Ключевые слова: De se, теория репрезентации дискурса, встраивание, фильм, точка зрения, ненадежный рассказчик
In the article "Unreliability and Point of View in Filmic Narration", Emar Maier investigates whether films such as Fight Club and Fear and Loathing in Las Vegas have unreliable narrators. He examines terminology of first-person and third-person narration, effaced or abstract narrators, and unreliability as they are applied to novels, and motivates a certain way of applying them to film. He argues that the films under discussion, and most films in general, have abstract narrators that are equivalent to an idealized camera position. Moreover, it is maintained that the notion of unreliability can sensibly be applied only to an agentive narrator who has information about some course of events, and thus can have inaccurate information, and/or can convey information about
38
© Dorit Abusch
A POSSIBLE-WORLDS CONSTRUAL...
the situation untruthfully or deceptively. In combination, these points indicate that filmic narrators cannot be unreliable. Instead, Maier proposes, passages that are described as unreliable involve embedding of film shots under implicit predicates that attribute the content of the shot to the perception, experience, or assertion of a character. In optimal terminology, then, these films have abstract narrators and make frequent use of embedding constructions.
It is illuminating to look at the argument with reference to possible worlds semantics, which is a model of information content that is applied in philosophy, linguistics, computer science, and economics. Any informational object such as a sentence, a novel, a picture, or a film has a "propositional content", which is modeled mathematically as a set of possible worlds. Informally, the propositional content is the set of worlds that the object describes accurately, e.g. the set of worlds of which the given sentence is a true description. In his 2011 dissertation, Gabriel Greenberg developed the application of this semantics to pictures, using formalized accounts of geometric projection to map pictures to possible worlds contents. The idea since then has been applied to more complex artifacts such as comics, film, and multi-media displays and narratives. My own work on pictorial narratives argued that the information content of a picture should be modeled as a relation between worlds and viewpoints, conceived of as relation that holds between a world and a viewpoint if and only if the world looks like the picture from the viewpoint [Abusch and Rooth, 2017; Rooth and Abusch, 2018]. The information content of a film is modeled in similar terms.
A first-person novel that uniformly represents the perception and informational perspective of a delusional or hallucinating character is in Maier's argument a true case of an unreliable narrator. Consider for instance a version of the novel Fear and Loathing in Las Vegas where all of the text is construed as representing the information and perspective of the drug-addled protagonist. Sentences such as (1), which describe huge bat-like creatures flying over the convertible occupied by the protagonist are typical. Let us stipulate that we are analyzing a grammatical form for the novel that has no hidden embedding operators, such as free indirect discourse operators. The syntax of the novel does include whatever hidden operators and/or indexing are responsible for temporal sequencing, anaphora, and the like.
(1) And suddenly there was a terrible roar all around us and the sky was full of what looked like huge bats, all swooping and screeching and diving around the car, which was going about a hundred miles an hour with the top down to Las Vegas. (Hunter Thompson, Fear and Loathing in Las Vegas.)
What is the propositional content of this novel, and in what way is the content unreliable? By virtue of sentences like (1) being unembedded,
DORIT ABUSCH
their meaning survives as an entailment of the novel as a whole, and any world in the propositional content has huge batlike creatures flying over a convertible. If the propositional content is unreliable, it is not because the proposition is false in a world that is an element of the content - by definition, any proposition is true in any world that is an element of it. Instead, I conjecture, readers infer a secondary propositional content, which in place of events of bats flying overhead in worlds in the content, have an event of an individual imagining bats flying overhead. The former "literal" content contains worlds that are radically abnormal by our standards, while the latter secondary content contains worlds which are normal by the standards of life in the United States around 1970. Unreliability is a matter of a disparity between the literal content and the secondary content.
Turning to film, Maier argues that, putting aside unusual cases such as films with a uniform point-of-view perspective, films have an impoverished narrator equivalent to an abstract camera location. Consider a slimmed-down version of the film Fear and Loathing in Las Vegas that uniformly presents the information and perception of the drug-addled protagonist. The film is shot from a neutral viewpoint, and so (Maier would argue) the narrator is impoverished. Such an impoverished narrator is not the kind of entity that can have information, and so a perception of unreliability cannot come from a conflict between the information of an agent and the content of the film. He would presumably analyze the entire film as embedded under an implicit "He sees...", or embed each shot of the film under such an operator.
I am skeptical of the necessity for this conclusion. I think that a viewer, presented with the film without embedding, and reasoning in a way that is mediated by the literal propositional content of the film, could infer that the film presents the information of a drug-addled individual who is depicted in most shots of the film. The viewer could infer a secondary content in the sense introduced above.
Bound up with these issues is the question of whether and how a neutral-viewpoint film shot, embedded or not, can convey agent-centered information. Here I think Maier possibly puts too much emphasis on the absence of first-person morphology in film, and is too committed to tying agent-centered information to the camera position. For the first point, what is conveyed with a third person pronoun in English in (2a) is conveyed in other natural languages including Amharic with a first person pronoun, with the equivalent of (2b) [Schlenker, 2003]. It is maintained that the embedded sentences in both kinds of languages convey information that is essentially agent-centered, for reasons that derive from a possible-worlds model of the information content of mental states. Lewis (1979) provides a theory that justifies the agent-centered paraphrase "Jack self-attributes the property of having a cold". This is de se interpretation.
(2) a. Jack believes he has a cold. b. Jack believes I have a cold.
A POSSIBLE-WORLDS CONSTRUAL...
These linguistic phenomena tend to suggest that rather than the opposition between first person and third person being the most important one, there is a central semantic and pragmatic notion of agent-centered information, that partially correlates with person morphology. Perhaps film is like the embedded context in (2a), in using agent-centered information, but not marking it with person morphology. In Abusch and Rooth (2022), we took up the problem of giving a possible worlds semantics for what Maier terms blended shots, such as the shot (3) from Fight Club, repeated from his paper. This shot shows the protagonist (on the right), and represents the information of the protagonist, because it shows an imagined individual on the left. Following Maier and Bimpikou (2019), such shots are embedded under predicates such as a dreaming predicate, an imagining predicate, or a hallucinating predicate, with a subject anaphoric to some discourse referent in the extensional context. In our analysis, inside the complement, the embedded panel or shot is incremented with a discourse referent that is interpreted as the de se parameter. This is linked up with the semantics of the embedded shot via the device of using geometric points or areas to introduce discourse referents for depicted individuals [Abusch, 2012; Abusch, 2020]. In (3), the de se parameter corresponds to the depiction of the character on the right in the picture, the character Jack. In this it is not a barrier that the de se agent does not have a geometric visual viewpoint matching the viewpoint for the film shot, because the discourse referent for the de se agent is introduced with a geometric discourse referent. In one technical version, this is a bounding box surrounding the depiction of Jack. Certainly, that the experiencer "takes the character on the right to be himself" is part of the intended interpretation of the film. This analysis puts information into the discourse representation that expresses this, and interprets the discourse representation in possible worlds semantics.
(3) Tyler Durden (Brad Pitt), left, and protagonist Jack (Edward Norton), right. Still from Fight Club (David Fincher, 20th Century Fox, 1999). Tyler is imagined by the experiencer Jack.
DORIT ABUSCH
In a film lacking embedding, such as the slimmed Fear and Loathing described above, it would in principle be feasible to use the same strategy, and introduce in the discourse representation the informational equivalent of a first person character (or more neutrally, a character with the role of experiencer), whose geometric viewpoint does not coincide with the viewpoint for the film. This adds to the surface appearance of the film and its basic semantics the information that a particular depicted individual is a counterpart of the experiencer.
One reaction to the account sketched above that refers to primary and secondary propositional contents is that while it provides an analysis of unreliability or the perception of it (or a plan for one), it is not an analysis that refers to narrators, their information, or the reliability of assertions they make. In this way the analysis is in agreement with Maier. There are no unreliable narrators in film. Instead, there are characters who are portrayed and whose information is portrayed (via embedding or not), with an implication that the information is not reliable.
References
Abusch, 2012 - Abusch, D. "Applying Discourse Semantics and Pragmatics to Co-reference in Picture Sequences", Proceedings of Sinn und Bedeutung 17, 2012.
Abusch and Rooth, 2017 - Abusch, D., and Rooth, M. "The Formal Semantics of free Perception in Pictorial Narratives", Proceedings of 21st Amsterdam Colloquium, 2017.
Abusch, 2020 - Abusch, D. "Possible-Worlds Semantics for Pictures", in: The Wiley Blackwell Companion to Semantics, 2020, pp. 1-31.
Abusch and Rooth, 2022 - Abusch, D., and Rooth, M. "Temporal and Intensional Pictorial Conflation", Proceedings of Sinn und Bedeutung 26, 2022.
Fincher, David. Fight Club. 20th Century Fox, 1999.
Gilliam, T. Fear and Loathing in Las Vegas. Universal Pictures, 1998.
Greenberg, 2011 - Greenberg, G. The Semiotic Spectrum. Ph.D. Dissertation, Rutgers, 2011.
Lewis, 1979 - Lewis, D. "Attitudes de dicto and de se", The Philosophical Review, 1979, vol. 88, no. 4, pp. 513-543.
Maier and Bimpikou, 2019 - Maier, E., and Bimpikou, S. "Shifting Perspectives in Pictorial Narratives", Proceedings of Sinn und Bedeutung 23, 2019.
Maier, 2022 - Maier, E. "Unreliability and Point of View in Filmic Narration", Epistemology & Philosophy of Science, 2022, vol. 59, no. 2, pp. 23-37.
Rooth and Abusch, 2018 - Rooth, M., and Abusch, D. "Picture Descriptions and Centered Content", Proceedings of Sinn und Bedeutung 21, 2018.
Schlenker, 2003 - Schlenker, P. "A Plea for Monsters", Linguistics and Philosophy, 1972, vol. 26, no. 1, pp. 29-120.
Thompson, H. Fear and Loathing in Las Vegas. Random House, 1972.
