CC BY
3
Эпистемология и философия науки 2022. Т. 59. № 2. С. 43-50 УДК 167.7
Epistemology & Philosophy of Science 2022, vol. 59, no. 2, pp. 43-50 DOI: https://doi.org/10.5840/eps202259219
Why philosophy of language is unreliable
for understanding unreliable filmic narration
Marc Champagne - PhD, Chair of the Department of Philosophy. Kwantlen Polytechnic University.
12666 72 Avenue, Surrey, BC V3W 2M8, Canada; e-mail: marc.champagne@ kpu.ca
A typical device in film is to have a character narrating what is going on (sometimes by voice-over), but this narration is not always a reliable guide to the events. According to Maier, distortions may be caused by the narrator's intent, naivety, use of drugs, and/or cognitive disorder/illness. What is common to these various causes, he argues, is the presence of a point of view, which appears in a movie as shots. While this perspective-based account of unreliability covers most cases, I unpack its methodological consequences and gesture at a possibility that Maier's analysis overlooks. A narration, I suggest, can be unreliable simply because it is ill-timed with the events shown on screen. In such a case, the distortion is not due to any character's point of view; rather, it comes from the film medium's ability to divorce what is seen and what is heard. As a consequence of this mismatch, it is possible to have a reliable narrator but an unreliable narration. Since voice and context of utterance usually match in ordinary speech, I conclude that philosophy of language may be ill-suited to properly understand this particular phenomenon. Keywords: film, narration, speech act theory, testimony, reliability, philosophy of language, subjectivity, semiotics
Почему философия языка ненадежна для
понимания ненадежного киноповествования
Марк Шэмпейн - доктор философии, заведующий кафедрой философии. Политехнический университет Квантлен. 12666 72 Avenue, Суррей, Британская Колумбия V3W 2M8, Канада; e-mail: marc.champagne@ kpu.ca
Персонаж, рассказывающий о происходящем (иногда голосом за кадром), - это типичный прием кинематографа, хотя такое повествование не всегда является надежным рассказом о событиях. Согласно Майеру, искажения могут быть вызваны намерением рассказчика, наивностью, психотропными веществами и/или когнитивным расстройством/болезнью. Объединяет эти причины, по его мнению, наличие точки зрения -перспективы, которая проявляется в кино некоторыми кадрами. Хотя объяснение ненадежности посредством указания на перспективу охватывает большинство случаев, я раскрываю его методологические следствия и указываю на возможность, которую анализ Майера упускает из виду. Я полагаю, что повествование может быть ненадежным просто потому, что оно не совпадает по времени с событиями, демонстрируемыми на экране. Искажение в этом случае не связано с точкой зрения какого-либо персонажа, скорее, это происходит из-за способности кинематографа разделять то, что видно, и то, что слышно. Вследствие такого расхождения можно иметь надежного рассказчика, но ненадежный рассказ. Поскольку в обычной речи голос и контекст высказывания обычно совпадают, я прихожу к выводу, что философия языка, возможно, плохо приспособлена для правильного понимания этого конкретного явления.
© Marc Champagne
43
Ключевые слова: фильм, повествование, теория речевых актов, свидетельство, надежность, философия языка, субъективность, семиотика
Throughout most of the movie Life of Pi, a narrator tells us what is going on. That narrator, we eventually find out, is in denial about the horrific events that happened. An earnest young Pi without the eventual trauma of resorting to cannibalism would presumably have told us the truth, but the perspective of retrospect gives us reason to doubt what the older Pi reports about zoo animals eating one another. As a philosopher of language, Emar Maier is concerned with understanding how such unreliable narration works. His conclusion, roughly put, is that films convey unreliability differently than novels do. Here is how I reconstruct Maier's argument:
(1) Novels that present fictional worlds must employ indicative statements.
(2) Indicative statements commit their speaker to their truth.
(3) Authors are not committed to the truth of their fictional works.
Therefore,
(4) There must be a fictional counterpart of the author who has access to the novel's fictional world and does the asserting.
(5) Movies tell stories by presenting a deliberate sequence of shots.
(6) A shot shows what the world is like.
Therefore,
(7) A shot must be taken as true by some subject.
(8) Filmmakers are not committed to the truth of their fictional works.
Therefore,
(9) There must be a fictional counterpart of the filmmaker who has access to the film's fictional world and sees events from their perspective.
(10) Seeing is a mental state.
(11) The medium of film does not have obvious analogues of mental state verbs.
Therefore,
(12) The narrator is the viewpoint(s) from which the film's fictional world is presented.
(13) Viewpoints are, by definition, subjective.
Therefore,
(14) Some viewpoints are unreliable.
(15) Making assertions and presenting shots from a given perspective are different things.
Therefore,
(16) "[M]ost if not all famous movie adaptations of novels with unreliable narrators are not themselves instances of unreliable filmic narration" [p. 35].
Much of this argument is beyond reproach. It is quite true, for instance, that in film one cannot rely as often on mental state verbs. It would be awkward and artificial to constantly say things like "She saw the handkerchief dangling from his pocket" (as per claim 11 above). Rather, one shows what she sees (in this case, the handkerchief dangling) and trusts the viewer to ascribe that subjective experience to the right vantage/character (claim 12). This technique is indeed quite different from novels, so Maier does an admirable job of advancing the debates. However, in what follows, I want to gesture at a possibility that Maier's inquiry overlooks. Specifically, I want to suggest that, in some instances, a narration's unreliability is best explained by the fact that we unreliably link together whatever sounds and images are given to us. Importantly, this mismatch - and the unreliability it gives rise to - is afforded by the filmic medium itself, not necessarily any character's point(s) of view. On my diagnosis, Maier's inquiry overlooks this possibility because of its emphasis on speech acts.
As Maier rightly says, in novels, "indicative statements" are used "for sketching a fictional world" [p. 26]. Context matters for fixing what statements are about. In a speech act like the marriage vow "I do," one says what one says when one says it. The same reliance on context applies when, for instance, I tell the waiter that the soup is too salty. Even a ventriloquist would be bound by stringent contextual constraints. Film, however, is edited, so the usual proximity constraints can be relaxed and toyed with, often to great effect. For instance, the cross-cutting near the end of Silence of the Lambs leads us to think that Jodie Foster's character is about to conduct a routine interview with a witness, when in fact she, not the SWAT team, is ringing the doorbell of the killer Buffalo Bill. Our habits must be added to the moving images and sounds for the sum to result in storytelling. We are duped (and enjoy an entertaining jolt of surprise) precisely because of our unquestioned desire to patch things together.
Just as we cannot help but connect shots into a more or less seamless narrative, so we cannot help but connect contemporaneous sounds and images. This is a fertile assumption, to be sure, since it is what we routinely witness in ordinary vision and speech. So, understandably, we carry that habit over to film and make the connection without any further thought. Expressed as a formula, the assumption that we implicitly rely on would be "same time = relevance." So, if a voice talks about a soup and a specific bowl of soup is shown, we conclude that the voice must be talking about that soup. Yet, given that this reasonable assumption is
neither mandatory nor always obeyed in film, it can sometimes generate unreliability.
Consider, for example, a situation where an otherwise reliable narration is shifted, such that it is heard, say, five minutes before or after the events it comments on. Prompted by the habit of uncritically assuming that the voice is always talking about what we now see, we would undoubtedly try to make the skewed combination make sense. Maier writes that "[u]nreliable narration occurs when the narrator presents a distorted view of the storyworld" and lists as typical causes of this unreliability youthful naivety, drugs, cognitive disorder or mental illness, and deliberate intent to deceive and/or lie [p. 26]. This helpful list may have a blind spot. Indeed, I want to suggest that, in some cases, we (in our filmic naivety) render the narration unreliable.
Maier briefly recognizes that "[s]hots are typically audiovisual depictions: the audio resembles (in some intuitive sense not to be made precise here) what the fictional world sounds like from some point (or path) in fictional space and time" [p. 27]. This captures well the standard assumption that viewers rely on when making sense of a film. However, precisely because it is an assumption, it a) hides from conscious scrutiny and b) admits of exceptions. One example would be the movie Primer, where the voice-over narration is a voicemail-like cassette recording. In terms of the taxonomy that Maier adopts, one can tell from the outset that this narrator is "personal," since he clearly has a stake in the events (although it takes a while to pinpoint who exactly recorded the message heard). In a way, this narrator is entirely reliable since, as befits a timetravel movie, he has advance knowledge of the events. In fact, his very first sentence, thirty seconds in, is "Here is what's going to happen." This lends support to the idea that "there must be some fictional entity, the narrator, that (i) has access to the fictional world and (ii) 'tells' a truthful story about it" [p. 28]. But, importantly, the narrator of Primer has no intention to deceive. Furthermore, he is not naive, not on drugs, and not mentally ill. Hence, whatever unreliability he displays stems from the bad timing between what is said (verbally) and what is shown (visually).
Of course, this counts as a mismatch only because we come to the film with our habitual assumption of good timing. The voice must be talking about what is going on, we reason. That inference is correct to a certain extent - the voice-over does bear upon the events as a whole. However, the skewed timing means that the voice-over does not always bear upon the specific events shown when the narrator is heard.
Maier's analysis comes in the vicinity of a case like Primer when it contemplates that "sometimes a narrating voice clearly does belong to one of the fictional characters, even if they are not shown as speaking in the shot and they are clearly narrating from a different place and time than what's shown in the shot. These voice-overs need not be reliable, they present what a specific fictional individual is saying or thinking and hence reflect their personal point
of view" [p. 6]. Maier develops his positive claim, which is that even when such a "first-person narrator can be considered fairly reliable, like Dr. Watson in the Sherlock Holmes novels, the events are presented as filtered through their consciousness" and can therefore be unreliable. Yet, while Maier tracks the subjective point of view of the narrator and inventories the various conditions that can render that point of view unreliable, he overlooks that our own assumption of relevant timing might be one of the sources of unreliability.
To better see this, consider the following thought-experiment. Even if God himself were to narrate with full truthfulness what is visually going on before one right now, God's narration could be rendered unreliable merely by decoupling His voice from the images at hand. This decoupling would not make God unreliable. Unlike the distorted perceptions in Fear and Loathing in Las Vegas, God sees things as they really are (from all angles, in fact). He is reliable, if any narrator is. This thought-experiment shows that even omniscient and well-meaning narrators are only one asynchrony away from becoming unreliable.
Maier omits music from his analysis, but playing a well-designed sound track at inappropriate times could generate similar interpretive unreliability. Alain Robbe-Grillet's L'homme qui ment offers an example of this. As the title announces, the man is a liar, but the movie abounds with departures from veridicality that no single character is responsible for. Nevertheless, our sense-making propensities cannot help but try to match what is seen and what is heard. So, when we see and hear a glass break seemingly at random, we place that event on a shelf, as it were, so as to later connect it to a glass that shatters audibly but not visually. This storybuilding on our part allows us to iron out the "audiovisual dissonance" [Chion, 1994, p. 38]. A less extreme example of mismatch between sound and image can be found in Hiroshima mon amour, in a flashback where a woman in a café recounts to a lover her past experiences. During the flashback, one hears the background music and noises of the café, while the images are from the past. Croaking frogs can also be heard, but those frogs are near the (present) waterside café, not anywhere in the (past) environments shown.
As Maier's contribution shows, many and perhaps most cases of unreliable narration can be fruitfully elucidated by tracking character viewpoints. Films like Life of Pi would fall into that category. But, I have claimed, this approach does not shed light on all cases. To achieve a dissociation of words and images, one needn't resort to narrator ill-intent, drugs, subjectivity, or a fancy time-travel premise. Rather, the two can come apart, simply in virtue of the medium. Films like L'homme qui ment fall into that category. The moment a film sunders our habitual expectation of match between narration and image, we become a source of unreliability, because we continue to blindly trust. Given our propensity for sense-making and default expectation of coherence, unreliability simply ensues. So, whereas Maier's "takeaway message [...] is that unreliable narration requires not just the presence of a narrator, but of a first-person
narrator" [p. 26], my more parsimonious takeaway message is that unreliable narration requires only a human possibility to err. The "first-person" in this case is us.
Like the trail of candies in Spielberg's E.T., good filmmakers feed us what they want in the sequence that they want to lead us where they want. We can mistakenly interpret signs, so filmmakers who can cleverly predict our reflexes (and obliviousness to them) can use errors in those reflexes to obtain their desired result. Yet, since viewers supply much of the meaning, a filmmaker is also free to become a "monstrator," distinguished from a "demonstrator" by exhibiting images and sounds without any overt narrative or didactic intent [Gaudreault, 2009, pp. 72-89]. Such a monstrative format is arguably as far as one can get from indicative statements and shots presented as true, so it needn't come into tension with the idea that filmmakers are not committed to the truth of their fictional works.
Maier mentions Banfield's [1982] suggestion that, in some cases (particularly in modern novels), events may "'narrate themselves,' without the mediation of an inferred speech act agent" [p. 25]. However, Maier decides to "put this narratorless option aside [...] and instead hold up the general speech-act-theoretic argument for the necessity of a narrator [...]" [p. 25]. Why put Banfield's suggestion aside? Following what he perceives to be a consensus in analytic philosophy of language [Pagin, 2016], Maier takes it as an axiom that "asserting presupposes an asserter." This is certainly reasonable when dealing with language. The problem, however, is that not every sign is linguistic or uttered with intent. It is entirely accurate to say that, in film, the "consumer of a piece of sequential media computes the basic discourse unit contents and the events they contribute and then infers as many plausible coherence relations between them as possible, given global rationality constraints like consistency" [p. 27]. I fail to see why this interpretive dynamic could not allow for events narrating themselves, sometimes unreliably so.
There can be many viewpoints to track in a film and Maier's analysis of unreliability clarifies much of what is going on. For example, hallucinations naturally seem veridical to whoever experiences them, so they are not always presented as hallucinations. Hence, even after the plot twist of Fight Club is revealed, viewers must keep in mind that, whenever they see Jack's fictional split-ego Tyler talking to Jack, Tyler cannot really be there. In Life of Pi, after the plot twist, the viewer is given a Jamesian choice to (retroactively) regard the animals as hallucinations or as real. In keeping with this complexity, I have gestured at some additional cases. Maier writes that "[o]nly when we detect, say, inexplicable weirdness or inconsistencies, or are given clues about drug abuse or mental illness, or a clear motive for lying, would we start reinterpreting the text as presented by an unreliable narrator" [p. 34]. However, examples from film history and armchair considerations about poorly-timed Godly reports
both establish that there can be a reliable narrator with an unreliable narration. The filmic medium itself, not skewed points of view, explains the possibility of such cases.
Famous movies like Life of Pi and Fight Club were adapted from novels. Maier holds that unreliable narration is "a misnomer" when describing those novels' film adaptations, since the on-screen versions actually rely on perspective changes instead of verbs "for attributing mental states to their characters" [p. 24]. I agree. For that reason, I think we need to switch from a speech-act-theoretic toolbox to a sign-theoretic one when moving from novels to films. Maier is aware that "[f]ilm shots convey their contents primarily iconically, while statements convey their contents primarily through grammatically structured conventional symbols" [p. 26], yet he falls in line with analytic orthodoxy by glossing pictorial content as "language-like" [p. 27]. To my mind, this unhelpful comparison with language renders Maier's inquiry insufficiently alert to the drastic conceptual changes that iconicity and indexicality often require. This shortcoming of language can be established on independent grounds, but Maier's work supplies us with an illustrative case study.
Speech act theory represents a significant achievement. But, given that there are all sorts of signs, a speech-based approach will invariably fail to do justice to some objects of study. Human culture, "that minuscule segment of nature" [Sebeok, 1985, p. 2] we feel most at home in, employs more than language, which represents only a part of the se-miotic spectrum. Indeed, "every form of communication is a solution to the same basic problem: how can privately held information be made publicly accessible through manipulations of the physical environment? Language is by far the best studied response to this challenge. But there are a diversity of non-linguistic strategies for representation with external signs as well [...]" [Greenberg, 2011, p. ii]. Mental states are particularly tricky to convey, which explains why the (verbal) devices used in novels to signal unreliability are not necessarily the (visual) devices used in movies. Yet, as Wittgenstein [1921] (1974) warned, the switch from telling to showing is a big one - bigger, at any rate, than the analytic tradition that he spawned often realizes. I thus submit that the possibility of a habit-rupturing audio-visual mismatch - along with the non-linguistic character of images and music - casts doubt on philosophy of language's aptness as a methodological lens through which to understand film.
I have provided original considerations to support this, but Maier's argument already establishes why philosophy of language is unreliable for understanding unreliable filmic narration. All that is needed is to draw that conclusion explicitly, as an additional claim inferred from Maier's claim 16. Luckily, the fix is not costly. Indeed, there is everything to gain and nothing to lose by switching to philosophy of signs or semiotics (as John Locke called it), since that approach already subsumes philosophy of language as a proper subset. So, if philosophy of language turns
out to have a blind spot for some cases, it is not anyone's fault. It's just that its perspective is, by nature, limited.
Acknowledgements
I want to thank Emar Maier for his stimulating piece, Ivan Mikirtumov (the guest editor) and Ahti-Veikko Pietarinen for inviting me to comment on it, and Martin Lefebvre for test-driving my initial intuitions in conversation.
References
Banfield, 1982 - Banfield, A. Unspeakable Sentences: Narration and Representation in the Language of Fiction. New York: Routledge and Kegan Paul, 1982.
Chion, 1994 - Chion, M. Audio-Vision: Sound on Screen. Trans. C. Gorbman. New York: Columbia University Press, 1994.
Gaudreault, 2009 - Gaudreault, A. From Plato to Lumière: Narration and Mon-stration in Literature and Cinema. Trans. T. Barnard. Toronto: University of Toronto Press, 2009.
Greenberg, 2011 - Greenberg, G.J. The Semiotic Spectrum. Dissertation, Rutgers University, 2011.
Maier, 2022 - Maier, E. "Unreliability and Point of View in Filmic Narration", Epistemology & Philosophy of Science, 2022, vol. 59, no. 2, pp. 23-37.
Pagin, 2016 - Pagin, P. "Assertion", in: E.N. Zalta (ed.). The Stanford Encyclopedia of Philosophy (Winter 2016) [https://plato.stanford.edu/archives/win2016/entries/ assertion/, accessed on 08.01.2022].
Sebeok, 1985 - Sebeok, T.A. "Vital Signs", The American Journal of Semiotics, 1985, vol. 3 (3), pp. 1-27.
Wittgenstein, [1921] 1974 - Wittgenstein, L. ([1921] 1974) Tractatus Logico-Philosophicus. Trans. D.F. Pears and B.F. McGuinness. London: Routledge, [1921] 1974.
