Café Lumière

An ‘innocent’ spectator viewing Hou Hsiao-hsien’s work might think they’re watching ‘nothing’ rather than witnessing the work of a master. Now an ‘innocent’ viewer in this context wouldn’t be an unsophisticated viewer: they may know and admire the intensely sophisticated work of Martin Scorsese and Alfred Hitchcock, love the virtuoso long-take crane shots of Brian De Palma and Robert Altman, but they would be innocent if they were unaware of Hou’s heterogeneous genius, his insistent need to keep the image full of unresolved possibility, full of possibilities that are not contrived narrative aporias leaving viewers to fill in gaps, but of the inexplicable, the inexplicability of a partial view on life. Let’s propose this is both an æsthetic and a philosophy.

Let us look at the former first, and a particular scene in Kôhî jikô (Café Lumière, 2003) where the young central character, Yôko (Yo Hitoto), goes and visits a friend who runs a Tokyo bookshop. In the scene, there is also the friend’s dog, and while there are points of focus within the scene – Yôko brings a watch back from Taiwan as a gift for Hajime Takeuchi (Tadanobu Asano) – Hou is more interested in the heterogeneous elements that make up the fabric of the sequence over the scene’s meaning. This would in the loose sense of the term put Hou in the tradition of Jean Renoir, Robert Altman and other filmmakers who concentrate the image, creating foreground and background density so that you feel there is always far more information in which to respond than can readily be absorbed. But filmmakers like Altman and Renoir are ‘busy’ filmmakers: part of their long-take sense of enquiry demands a constant hubbub of activity. We might think of an early scene in Renoir’s La Régle du jeu (1939) when guests arrive and the director has characters coming in and out of the frame in all directions. Meanwhile, critic Pauline Kael astutely referred to Altman’s A Wedding (1978) as “like a busted bag of marbles – people are running every way at once” (1).

In Hou’s work there is a hint of the Altmanesque and the Renoiresque, but the sensibility is very different: his is certainly a full image, but a sedate as opposed to hectic one. In the scene with Yo, Hajime and the dog, we have a number of things going on and yet we have no sense of a burst bag of marbles. Hou works a paradox: he fills his frame but retains a sense of stillness. By often framing in such a way that he keeps human motivation and activity vague, and also by making objects and the non-human (like Hajime’s dog) visually as central to the scene as his subjects, Hou can offer rarification within the full scene. We might think here of sequences in Nanguo shang hua (Goodbye South, Goodbye, 1996) where Hou allows a fight to take place in long shot, or a suicide in Hai shang hua (Flowers of Shanghai, 1998) to take place at the rear of a scene – so much so that we can easily miss the moment when the character jumps out of a window and kills herself. By positioning his camera in such a way, or by suggesting we should pay more attention to the foreground than the background, Hou can achieve the obscure, the inexplicable, the uncanny not in subject matter but in the very form.

In subject matter, Café Lumière could almost be a Yasujiro Ozu film (maybe even, to draw comparisons with a Taiwanese contemporary, an Edward Yang film). It was of course made in homage to Ozu, made to celebrate the centenary of Ozu’s birth, but Ozu, one of the masters of film, is nevertheless a master of convention. His work frequently moves towards the comprehension of the meaningfulness of family, the importance of friendship, the significance of marriage. As Donald Richie says,

the life with which Ozu is concerned in so many of his films, then, is traditional Japanese bourgeois life. It is a life singularly lacking in the more dramatic heights and depths found in society less conspicuously restrained. (2)

Richie goes on to quote G. K. Chesterton: “Tradition means giving votes to that obscurist of classes, our ancestors. It is the democracy of the dead.” (3) Ozu, then, is a democratic traditionalist and though it’s been said Hou’s personal position shares similarities with Ozu’s – that he has problems with the younger generation who lack a respect for tradition – this is often not part of his æsthetic focus. He dwells here, in Qianxi manbo (Millennium Mambo, 2001) and Goodbye South Goodbye, for example, not so much on tradition as on the characters’ restless energy. Whether this takes the form of the fights in Goodbye South Goodbye, the throbbing techno music in Millennium Mambo or the central character in Café Lumière refusing to marry the Taiwanese boyfriend she’s been impregnated by, Hou shows characters adrift and searching. They are usually unconventional and, though Hou’s style seems in many ways as still as Ozu’s, Ozu’s stillness is activated by dialogue, as the characters slowly move around apartments, sip saki or sit at an office desk. Hou’s thematic is more likely to be activated by a camera movement, or use of the full width and depth of the frame as he finds a position at one remove from the restlessness of his characters.

We can see, from the perspective we’ve proposed, that critic Mark Cousins was generally right to insist in his book, The Story of Film, that Ozu should be the centre of cinema, not Hollywood. As he says

if Ozu’s people are most clearly at the spatial centre of his films than they are in any other filmmaker’s work, and if his world view – his sense of the possibility of social and psychological change – is so measured, then in this sense at least, his body of work, more than any other, is the centre of the map. (4)

Hou, for all his admiration of Ozu, would seem to be searching out much more the peripheral than the central, in all senses of the term. His approach decentres cinema as he moves towards making film not a space for observation of the human, but a space in which the peripheral can become central: it becomes central by virtue not of its positioning within the frame (the way Altman sacrifices composition to search out the peripheral) but the viewer’s acute sense of looking out for the peripheral observation, as we scrutinize the carefully composed composition trying to focus on what exactly should matter. Thus, it is space for human observation, not first and foremost the observation of the human.

Café Lumière

Hou has thus taken on board the Bazinian idea of the de-dramatized action, where the simplicity of observing behaviour destroys drama, and yet takes it further by turning it into the inexplicable de-dramatized action. So when André Bazin believed the maid making coffee destroyed cinematic storytelling, he was still holding to a degree of epistemological certitude, no matter if the images’ “ontological equality destroys drama at its basis” (5). In a Hou film, we wouldn’t necessarily be sure what the maid would be doing, and this is interesting, because, out of the obscuring of information, Hou achieves a greater degree of cinematic tension than Bazin’s position would offer. He doesn’t quite return to the narrative that Bazin proposed was being destroyed by focusing on the minutiæ, but he does give cinema once again a type of suspense. It’s a suspense not in the narrative, but in the perceptual viewing process, in withholding things we would like to see, and leaving us caught between expectation and frustration. These elements are, after all, the building blocks of suspense cinema, but we expect the expectation to be satisfied and frustration alleviated. So a filmmaker like Hitchcock, say, would hide the face of the murderer who is being chased by the cop. At the end of the chase sequence, we’d expect the face to be revealed, or, if the baddie escapes, we will be in a state of anticipation waiting for the face to be revealed at a later stage of the film – when he is finally caught. The expectation/frustration gap will be worked manipulatively, and any information withheld will eventually be offered because Hitchcock sets up an expectation that demands the viewer will be satisfied. In Hou’s work, there is no such expectation, so the frustration is much more phenomenologically immediate. Early in the film, there is a scene where the young female character talks to her neighbour out of frame and we would like, as we’ve come to expect, to know who it is who is just outside of the shot. But Hou eschews the shot, only to introduce us to her ‘properly’ much later in the film when we would long since have lost any interest in who that person off screen was.

Thus, where Hitchcock eschews for suspense, it would seem Hou eschews almost out of indifference. But this wouldn’t be fair. Hou instead wants us to be expectant within the shot and thus to put all our curiosity into the immediate sequence, rather than us constantly thinking of how it fits into the full narrative of the film. He wants us to direct our look at what we think matters, and then to extend that gaze beyond what we can see within the frame and feel the frustration of a partial view, but not expect us to be rewarded later for that sense of frustration. Thus, late in the film, when we see the neighbour at last, if we have any response to her presence it will not be the sort of narratively driven sense of relief, but perhaps a novelistic sense of curiosity, the curious surprise we might feel when a favourite novel is adapted and we’re surprised at the way the person looks, because we had imagined them differently.

What we’re proposing, then, is that Hou’s is a great cinema of the imagination. This is imagination not in the sense of the realised imaginary – where the film imagines for us, as in a sci-fi film with its wonderful set design, or an action film with its vividly choreographed chase sequences – but, instead, where the filmmaker forces us to use our imagination in a way not entirely dissimilar to the novelist, when a writer, if by necessity, gives us only so much information on a character and we feel we have to do the rest of the work ourselves. If a character in a novel is described as beautiful, no matter how vivid the description of that beauty, we have to make a leap that allows us to believe that beauty in the way a film can call a character beautiful, show us the character and we can judge that beauty. Hou, though obviously working in film, is finally a filmmaker who wants us to believe rather than to judge. He wants to use off-screen space and the active but fairly neutral on-screen space in such a way that we are in a constant hermeneutic with the image; we’re constantly querying what he’s showing us, why he’s showing it to us and what we should be prioritising in what we see.

In an essay on “Empathy in the Fiction Film”, Alex Neill quotes Aristotle concerning our imaginative and emotional faculties in relation to art. Aristotle, he explains,

holds that the pleasure that we take in mimetic works is a pleasure that comes from learning. The source of the pleasure that we take in tragedy, he holds, lies in the arousal and subsequent catharsis of pity and fear. (6).

Frequently in Hou, however, catharsis is rarely possible because the ‘learning’ is too ambiguously presented to allow for the certitude of an emotional reaction. If Bazin suggests the destruction of drama, he nevertheless isn’t proposing the collapse of an emotional response to the image. When Bazin says

what kind of causal relationship could you establish between a harmless angina for which Umberto D [Carlo Battisti] will be treated in hospital, his land-lady’s turning him out onto the street and his thinking of suicide (7)

Bazin reckons the director avoids a sentimental causal logic here, and that is a fair point, but there is still an unequivocal emotional through-line that leaves us strongly attached to the titular character. This emotional through-line is much harder to find in Hou and may be usefully explained, and explored, through a further Neill comment, when he invokes Nicolas Roeg’s Don’t Look Now (1973) and our empathetic relationship with film fiction. As Neill proposes, we have far more empathic engagement with film characters than we do with literary ones:

of the people with whom I have discussed this issue, almost all make some sort of appeal to identification and/or empathy in characterizing their responses to movies; far fewer do so in characterizing their engagement with literary fiction. And this tendency is mirrored in theoretical writing about our engagement with fiction. (8)

Café Lumière

But what Hou is doing – and he is not alone in this, though he might be pushing this aspect further than any other contemporary filmmaker – is asking for the imaginative leap we’ve already proposed that is expected of fiction and attached it to an eschewal of conventional cinematic empathy, an empathy so easily achievable in cinema. If Neill says that one reason we don’t empathise with literary characters to the degree that we do in film is because we know too much about them, then Hou wants us neither to identify nor sympathise: we know too little to comprehend them, and we don’t have the easy comprehension of their visual world with which to place them. They do not become steadily more knowable to us as the film continues, but if anything more inexplicable. When Hou places his central character in Café Lumière in the Tokyo milieu, he does so not to establish her character, but if anything to dissolve it into the city. When Neill says “our motive for empathizing with others, I suggest (and it may not be a conscious motive) is the desire to understand how things are with them” (9), we could say Hou’s purpose is to make us take a further leap into their world because their world is held at arm’s length from the viewer. If usually we expect from film fiction a world vividly described because the film has the visual means at its disposal with which to describe that world, then in Hou’s work this is a luxury he deploys very frugally.

It is a perfectly valid question to ask why, and we might wonder whether this is simply an avant-garde vein, a vein described by Holger Roëmers in an interview with another contemporary Asian filmmaker, Apichatpong Weerasethakul, who made Sud Pralad (Tropical Malady, 2004). Here he suggests avant-gardists regard deviations

from traditional forms as proddings to make the spectator ‘work’; but the allure of [Weerasethakul’s] Blissfully Yours [Sud sanaeha, 2002] resides in its blithe disregard for this sort of work ethic and its insistence that viewers play along with its puckish defiance of narrative propriety. (10)

Hou may be compared in the interviewer’s introduction to Weerasethakul, but Hou is much more interested in this work ethic. This is more or less what Robin Wood pinpoints when, at the end of his article on Flowers of Shanghai in CineAction, he believed that the film’s ‘scheme’ was “in fact so subtly worked that it has taken me at least six complete viewings […] to disentangle it” (11). We needn’t give a Hou film that degree of close analysis, but by the same token it would seem closer to mimesis, to the learning curve proposed by Aristotle, than immediate sensation, or æsthesis. Mimesis and æsthesis are terms used by Jacques Ranciére in a recent book, reviewed by Tony Wood in New Left Review. Here there is

an ambition to supersede conventional representation for direct sensory affect – displacing mimesis by aesthesis – was inscribed in it from the start. At work here was what Ranciére has described as the ‘aesthetic revolution’, when stable codes of artistic expression were overturned and matter released from its subordination to formal constraints. (12)

Here Ranciére proposes that a useful separation can be made between an art that works first and foremost on the nervous system, an art that has become increasingly present in the 20th Century, to a long tradition of mimetic art that wants to capture reality formally – to frame it in a comprehensible way, whether that be Aristotle’s notion of unity of time and place in drama, or classical rectangular framing in art. If a critic like Bazin wanted to push mimesis much further, he did so by suggesting, in his above quote for example, the way that drama could be undermined and ‘reality’ fore-grounded. But, of course, Hou pushes this further still, insisting on mimesis, certainly, but a mimetic ambiguity that will use a series of devices that undermines our sense of certitude in what we witness. We can see he respects the mimetic, but without Aristotle’s formal unification or Bazin’s epistemological purposefulness. Thus, Hou’s mimetic ambiguity goes further than Bazin’s “ontological ambiguity of reality”, because, though Bazin is a masterful critic of the ambiguous, it is an ambiguity contained by a loosely theological perspective, evidenced in such a comment as Bazin’s “take a close look at the world, keep on doing so, and in the end it will lay bare for you all its cruelty and ugliness.” (13)

Could Hou make such a statement, a statement relevant to the work of loosely socialist realist filmmakers like early Roberto Rossellini, Vittorio de Sica and, more recent, Ken Loach? One doubts it, and this might be a useful juncture at which to differentiate between Bazin’s epistemological purposefulness within ontological ambiguity and Hou’s epistemological troublesomeness within the apparently same ontological ambiguity. In Bazin, we have a world that is ontologically ambiguous; but the filmmaker searches out truths within that ontological ambiguity. In Hou, there is a far stronger sense of truths being hard to find. Even if the filmmaker looks hard enough, and few filmmakers look harder than Hou, will cruelty and ugliness reveal themselves? Let us propose for example that Hou does have a problem with the younger generation’s aimlessness. Then what about the scene near the end of Café Lumière where the mother, the daughter and the father eat together after the parents come and visit their daughter in Tokyo? When the mother leaves the room, there is the perfect opportunity for the father and daughter to confront each other. But when the father looks at the daughter, the daughter looks away, and when the daughter looks at the father the father looks away. A filmmaker who believed, as Rossellini did in his earlier work, that he “was retaining a moral position more than a style” would, if he had a problem with the younger generation, use such a moment to make a moral point: the daughter could not confront the father whose values she cannot share but wishes she could. Instead, both characters are equally incapable of communicating and any ideological point concerning the generations gives way to a stylistic point. If Rossellini could say we were pursuing a moral position more than a style, in Hou we can reverse it: the style is the moral choice. It is finally more important to respect the ultimate ambiguity of the image than to move towards ethical purposefulness.

This is an approach Hou shares with a number of other Asian filmmakers, including Tsai Ming-Liang, Jia Zhangke and Hirokazu Kore-eda. This type of ambiguity is not only very different from that proposed by Bazin, but also very different from that proposed by Alain Bergala on a fine essay on Godard called “The Other Side of the Bouquet”. Here Bergala proposes that generally films are shot from the most appropriate place:

In most films, where the authors try to efface themselves beneath the story they’re telling, the attack always takes cues from the composition. The common-sense approach determines the vector of the shot’s readability, and affirms the good faith contract with the spectator. (14)

He sees, however, that Jean-Luc Godard breaks with this approach, shooting from the reverse angle of our expectation. As Bergala says, telling an anecdote about Auguste Renoir in conversation with Henri Matisse, “I often paint bouquets from the side I haven’t arranged.” (15) This approach we can also see very clearly in Robert Bresson and perhaps in Wong Kar-wai, as we feel they don’t so much set the scene and film it, but reverse the expected angle and film it, and this gives them either freeeplay within the mise en scène, or their very own anti-theatrical rigour (Bresson most obviously). We can’t be sure where they will shoot from next. They take full advantage of the cinematic potential in the scene. As Bergala suggests,

one of the filmmaker’s pleasures, however – a pleasure unknown to the theatre director – is that his art, like the painter’s, permits him to treat composition and attack separately. (16)

Café Lumière

Hou does not generally do this, however. He’s more likely to work from the frontal perspective we might expect from a stage play, and then move the camera according to the original set-up, either relying on the camera to move around a 180-degree axis, or, if using a fixed frame, to bring the actors in and out of the filmic space. What he’s unlikely to do, and this is partly what makes Hou, like Jia Zhangke, Tsai-Ming Liang and Kore-eda, a long-take filmmaker, in the way Godard, Bresson and Wong Kar-wai are montage filmmakers, is cut to the reverse angle. We’re unlikely to be given a sense of privilege that cutting can provide, but neither will we be especially disorientated in the cutting – as we might be in a Wong Kar-wai film, or for that matter in a Dogme film, where the constant reverse angles throw our sense of orientation.

This is, of course, consistent with Bazin’s demand for the long-take over editing, but it would be wrong for us to assume Hou’s purpose is exactly the same as Bazin’s, because, as we’ve proposed above, what counts is not the clarity of the world’s moral truth, but its deeper impenetrability. Let us take for example a scene from Goodbye South Goodbye, where there is a tense exchange a period of time after one of the characters has attempted suicide. At a key moment, one of the characters asks another character in the scene whether he knew about the money the suicidal character owed and, as he asks, we would expect a cut to the other character’s reaction to the question. Instead, Hou holds the image and the character who asks completely blocks out of view the character who is being asked, so that we have no idea except from his voice how he’s responding to the question. Though ostensibly Hou is telling a story about ‘primary’ motives, he tells it in, if you like, a ‘secondary’ way. The characters may function chiefly off greed, lust and ambition as Hou tells a gangster story, but he films it so obliquely that the motives within the wider motivations become hidden. Hence we can’t often tell how a character is responding to another character because Hou will film motivated actions without motivated camera movement and montage. Shortly after the questioning, the questioner goes over and starts hitting the one questioned, and, if we don’t see the violence coming, it is in part because the violence has been cinematically unmotivated. If we take into account Bergala’s comments about appropriateness and good faith contracts, we would expect a series of shot-reverse-shots between the questioner and the one being questioned, and as the scene becomes more fraught with tension, we might expect a series of shots moving into close-ups so that we see the anger on the characters’ faces. Hou, though, has the questioner facing the one questioned but with his back to the audience and, in the process, obliterating the other character. The motivations of the characters are being ‘undermined’ by the de-motivation of the camera.

Why does Hou do this? There are several reasons, perhaps. But one may lie in the fact that recognition of aggressive primary motivations are very familiar to us from the cinema but very uncommon in our lives. It’s that one about bring a gun into a film and you’ve suddenly got a story. But, at the same time, in ‘real life’ we don’t just have a story: when a gun is brought into a scene, we have a ‘situation’. Generally mainstream cinema turns situations into stories, turns our worst nightmares in life into wish-fulfilment narratives. We want something to happen and cinematically it does, even though, of course, in life we would very much prefer it didn’t happen. Bergala may talk about the good faith of the contract, but does this good faith demand on the viewer’s part a bad faith – a desire from entertainment what he doesn’t desire from his life? Does Hou ignore the contract and demand, instead of a contractual good faith, an existential good faith, the good faith of inexplicability in relation to situations, rather than the explicability of storytelling? Thus, Hou films situations rather than stories and this leads back to an earlier point while expanding our present one. For Hou, morality isn’t strictly a question of tracking shots, but it is a question of attack, or of limiting the attack once the mise en scène had been set up. It is as though he has his own 180-degree rule that insists on keeping his scenes situational and refusing the story, and refusing the story so strongly that if we enter into the emotional lives of the characters we do so at a situational remove instead of on the grounds of narrative involvement.

If, then, we are to have an empathetic relationship in a Hou film, taking into account Neill’s comments, we will find ourselves involved much more in the situation than the story, and we could argue that this is finally where profound empathy lies. It doesn’t lie in a certain type of identification, as Neill proposes, but in an ongoing process of comprehending situations, of paying attention to the complexity of situations. This is Hou pushing Bazin further. Sure we must look closely at the world and keep on doing so, and, if it will not show all its cruelty, then at least its multiplicit complexity will become clear to us. It will take not the glance of storytelling, but the gaze of observation. There is a nice comment from Nathan Lee in a recent issue of Film Comment where he says,

good art makes you notice things; great art alters the way you look at them. The difference is a matter of attention, concentration. Great artists don’t discover anything new, they just look longer and harder at what’s in front of them than anyone else […] (17)

It is as if Hou’s own 180-degree rule focuses his attempt to look longer and harder than just about any filmmaker before him. It wouldn’t be immediately obvious that Hou is in the Bazinian tradition, but from a certain angle, so to speak, he certainly is. He just pursues it with a rigour perhaps previously unmatched, and assumes no underlying moral truth.


  1. Pauline Kael, When the Lights Go Down (New York: Holt, Rinehart and Winston, 1980), p. 441.
  2. John Orr and Olga Taxidou (Eds), Post-war Cinema and Modernity: A Film Reader (Edinburgh: Edinburgh University Press, 2000), p. 269.
  3. Orr and Taxidou, p. 269.
  4. Mark Cousins, The Story of Film (London: Pavilion, 2004), p. 129.
  5. André Bazin, translated by Hugh Gray, What is Cinema? Vol II (California: University of California Press, 1972), p. 81.
  6. David Bordwell and Noel Carroll (Eds), Post Theory: Reconstructing Film Studies (Wisconsin: University of Wisconsin Press, 1996), p. 179.
  7. Bazin, p. 80.
  8. Bordwell and Carroll, p. 187.
  9. Ibid, p. 188.
  10. Holger Roëmers, “Creating His Own Cinematic Language: An Interview with Apichatpong Weerasethakul”, Cineaste, Vol XXX, No. 4.
  11. Robin Wood, “Flowers of Shanghai”, CineAction, Issue 56, pp. 11-9.
  12. Tony Wood, “The Ecstatic Spiral”, New Left Review, November-December 2002, pp. 141-8.
  13. André Bazin, What is Cinema? Vol. 1 (California: University of California Press), p. 27.
  14. Raymond Ballour and Mary Lea Bandy (Eds), Jean-Luc Godard: Son + Image (New York: The Museum of Modern Art, 1992), p. 57.
  15. Ballour and Bandy, p. 57.
  16. Ibid, p. 57.
  17. Nathan Lee, “Pleasures of the Text”, Film Comment, September-October 2005, p. 17.