Noel Burch

To the
Distant Observer
Form and meaning
in the Japanese
cmema
TO THE DISTANT OBSERVER
To the distant observer
They are chatting of the blossoms
Yet in spite of appearances
Deep in their hearts
They are thinking very different thoughts
KI NO TSURA YAKI
 Noel Burch

To the Distant Observer

Form and Meaning in the Japanese Cinema

Revised and edited by Annette Michelson

University of California Press

Berkeley and Los Angeles
University of California Press
Berkeley and Los Angeles, California

Copyright © Noel Burch 1979

Film set in ro/ 12 point Times

Printed and bound in Great Britain
by W & J Mackay Limited, Chatham
Contents

Foreword and Acknowledgements 7

Preface I I
Some Terminological Indications I8

Part I. Grounds, Premises

I A System of Contradictions 25
2 A System of Signs 35
3 A Boundless Text 42

Part 2. A Frozen Stream?

4 A Machine Appears 57
5 A Parenthesis on Film History 6I
6 A Rule and its Ubiquity 67
7 Bulwarks of Tradition 75

Part 3· Cross-Currents
8 Transformational Modules 89
9 Lines and Spaces 93
IO The Fate of Alien Modes roo
II Displacements and Condensations I IO
I2 Surface and Depth r 17
13 Kinugasa Teinosuke I23

Part 4· Iron Trees, Golden Flowers

14 ·The Weight of History and Technology 141
15 Some Remarks on the Genre Syndrome 151
r6 Ozu Yasujiro 154
I7 N aruse Mikio and Yamanaka Sadao r 86
I8 On Architecture 198
19 Ishida Tamizo 202
20 Mizoguchi Kenji 2 I7
21 Shimizu Hiroshi and Some Others 247
22 Epilogue to a Golden Age 262

Part 5· A Chain is Broken

23 Film and 'Democracy' 271
24 Kurosawa Akira 291
 Part 6. Post-Scriptum
25 Oshima Nagisa 325
26 Independence: its Rewards and Penalties 345

List of Works Consulted 365

Appendix I: Check-list of films by Ozu, Mizoguchi,

Kurosawa and Oshima 368

Appendix 2: Archive holdings in Japan: Japanese

films produced before I946 372

Publisher's Acknowledgements 377

Index of Films 378

Index of Names 385

NOTE
Illustrations are numbered Figs. I-66. Some figures
consist of sequences of shot analyses taken from the
frame; the individual shots are numbered Pl. I, 2, 3, etc.
Frames from the same shot are numbered
e.g. Pl. Ia, Ib, IC.
Foreword and
Acknowledgements

All the Japanese names are written in the standard Japan-

ese form: family name before given name. The single
exception is that of the Americanized actor, Sessue
Hayakawa.
Titles of Japanese films are given in English with the
Japanese title and the production date at the head of the
main entry. I have tried to render them faithfully. My trans-
lation does not always correspond to those given by other
sources particularly in the case of films unreleased in
English-speaking countries. Whenever films have one or
more English release titles, these are given in a footnote to
the main entry. Thus, a well-known film may often be
found in the index under three or more titles.
Unless otherwise indicated all translations from French
are mine.

This book could never have been written without the help
of many friends in Japan and elsewhere. First of all, I must
thank Shibata Hayao, Kawakita Kasuko and the entire
staff of the Shibata Organization for having made possible
my first trip to Japan and for having done so much to
facilitate what is inevitably a difficult acclimatization. On
my second trip, it was my dear friends Kuraoka Akiko and
Philippe Ferrand who helped to make bearable the gruel-
ling rhythm of Tokyo life.
Special thanks are due to Ohba Masatoshi of the Tokyo
Museum of Modern Art Film Centre whose devoted assis-
tance made it possible for me to make shot-by-shot trans-
criptions of dozens of films and to examine scores of others.
I am also greatly indebted to Kawakita Kashiko and Shim-
izu Akira of the Japan Film Library Council, who arranged
for me to study many films from their collection and who
helped me in so many other ways. Credit in particular is due
to the Council for all the separate film stills illustrating this
volume, as well as the sequence from A Tale of Late
Chrysanthemums. And I especially wish to thank Ema
Michio of the Kyoto Film Library for entrusting me with
FOREWORD AND some twenty16mm prints and a projector. I was thus able
ACKNOWLEDGEMENTS to screen those rare films in my hotel room when I unwit-
tingly arrived in Kyoto on the eve of a holiday. Matsuda
Shunji graciously made available to me even rarer films
from his collection, while Hiroko Govaers gave me invalu-
able practical and linguistic help in both Tokyo and Paris.
My talks with Donald Richie and Yoshida Tieo were
extremely helpful in clearing up certain points of film-
history. The critics Kawarabata Nei and especially Sato
Tadao gave me much valuable guidance. I also wish to
thank Iwamoto Kenji for his stimulating theoretical
insights and James Leahy, Dr Barry Salt, Jorge Dana and
P. Adams Sitney for their suggestions and criticisms.
However, my greatest debt is to Annette Michelson. As
the title page suggests, without her severe editing and critic-
ism, as well as her helpful suggestions, this book, my first in
my native language, could never have been made readable
or, indeed, comprehensible.
N.B.
FOR MARTINE

The point is not to seek truth or salvation in the

pre-scientific or the philosophically pre-conscious, nor to
transfer whole segments of mythology into our
philosophy; in dealing with these variants of mankind
who are so different from us, our aim should be to gain
further insight into the theoretical and practical problems
which confront our own institutions, to gain new aware-
ness of the plane of existence in which they originated
and which the long record of their achievements has
made us forget. The 'puerility' of the East has something
to teach us, if only the narrowness of our adult ideas.

Maurice Merleau-Ponty: Signes, p. 175

Preface

Let it be clear from the outset that this book lays no

claim to being yet another 'history of the Japanese
cinema'. In the native language there already exists
Tanaka Junichiro's four-volume factual compendium; 1 in
English, it is hoped that the reader will be familiar with
the several writings of Donald Richie and in particular
The Japanese Film, written in collaboration with Joseph
Anderson and drawing much data from Tanaka; in
French, a forthcoming volume by Max Tessier will dis-
play, I understand, similarly comprehensive ambitions.
Mine are at once more modest and, considering the
limited tools at my command and the material at my dis-
posal, considerably more ambitious, not to say presump-
tuous. My approach is, of course, historical in every
sense. In fact Japanese cinema from the period 1917-45,
to which the bulk of this essay is devoted, is likely to
seem quite remote to both Western and Japanese read-
ers. My reading of the films is conducted, moreover, with
constant reference to the history of 'Japanese culture' as
a whole. For it is beyond doubt that Japan's singular his-
tory, informed by a unique combination of forces and
circumstances, has produced a cinema which is in essence
unlike that of any other nation.
This essential difference between the dominant modes
of Western and Japanese cinema is the main concern of
this study. It is intended, furthermore, as a step in the
direction of a critical analysis of the ideologically and cul-
turally determined system of representation from which
the film industries of Hollywood and elsewhere derive
their power and profit. It consequently may be under-
stood as part of a much broader movement, that of the
modern search for a Marxist approach to art, initiated by
Brecht and Eisenstein, and which involves a detour
through the East.
Paradoxically, it is in France that the scrutiny of le texte
de !'Orient has been most productive. Paradoxically, r. Nihon eiga hotatsu, Chuo Koran-sha,
because oriental scholarship per se in post-war France Tokyo, 1957.
PREFACE has been scant and on the whole mediocre. The writers
associated with the review Tel Que!, 2 who have been at
the centre of this new concern with the East, have had to
rely almost exclusively on the pioneer work of Marcel
Granet, La Pensee chinoise ( 1934), or on more recent
wntmgs in English (in particular those of Joseph
Needham and his team). Understandably, the writings of
Philippe Sollers, Julia Kristeva and their colleagues have
tended, in so far as they have touched on the Orient, to
stress the implications for contemporary developments in
the theory of dialectical materialism of the thought-
systems of ancient China. They have emphasized, in par-
ticular, the importance of non-phonetic writing in the
development of modes of thought and social practice
antithetical to those characteristic of the West within the
structure of capitalism and its Aristotelian and Christian
heritage. Comparatively little attention has been paid to
Japan by these authors, partly no doubt because modern
Japanese history would seem to indicate, on the contrary,
remarkable compatibility between 'Japanese thought pat-
terns' and the ideological superstructures of capitalism.
Japanese artistic and social practice have, moreover,
produced almost no theoretical practice comparable to
the logic and linguistics of ancient India or the cosmology
and science of China.
The fate of Buddhist logic, subsequent to its importation
into Japan as part of the body of Buddhist teachings,
typifies the Japanese disdain of theoretical practice. This
highly refined discipline of Indian origin rapidly became a
form of ritualized debate which
continues to be held at Mt Koya, even to this day. In this ritual, the
answerers (rissha), the questioners (monja), the judge (tandai),
the stenographer (chuki), and the manager (gyi5ji) sit in pious
attitude around the statue of the Buddha according to fixed rule.
Buddhist hymns are sung and siitras are read. Thus, in Japan,
logical debate was reduced to a mere Buddhist meeting, a
decorum of the most pious form. Further, the form ofthe ritual was
extended without change to the poetic debate or utaawase. 3
The specific traits of Japanese 'theory' must be sought, not
in any body of theoretical writings, ancient or modern, of
2. Advanced literary review which in
the late 19005 and early 19705 played an
Japanese origin, but in the practice of her arts and letters. It
important role in the development of a is significant in this respect that the only rudimentary
seminal but often debatablv
metaphorical theorization ~f literary
attempts at aesthetic theory, indeed theory of any sort, are
history and practice grounded in to be found in the incidental writings of such figures as
structurallingui5tics, psychoanalysis
and historical-dialectical materialism.
Tsurayaki, the Heian poet and anthologist, Zeami, actor/
author and codifier of the no play, Chikamatsu, the
3· Nakamura Hajime, The Ways of
Thinking of Eastern Peoples, p. 545·
eighteenth-century dramatist, etc. 4 The larger theoretical
(N.B. The italics arc mine.) implications of Japanese practices are to be derived

12
through a reading conducted from outside the culture PREFACE
which has produced them, for the very notion of theory is
alien to Japan; it is considered a property of Europe and the
West.
One of the principal assumptions of this study is that the
critical framework developed in France over the past
decade (partly through an investigation of Eastern
thought) provides elements towards an understanding of
the far-reaching theoretical implications of le texte
japonais. For, just as it is possible to read a written text as
complex system, both self-contained and historically
determined, so too with a culture- and Japanese culture
perhaps more relevantly than any other. I like to think that
it was this clarifying vocation of French thinking in general
which a Japanese scholar, Toki Zemmaro, had in view
when he remarked to the French Japanologist Rene Sief-
fert that 'French was the only language precise enough to
render with precision the full imprecision of the no play'. 5
Of course, this 'imprecision' of the language and 'thought-
patterns' of Japan is another cliche, belaboured by
Western-centred scholars but also by many Japanese,
impressed by the marvellous compatibility which they
observe between the highly 'rational' linguistic and theore-
tical practices of the West and the technological tasks
involved in the construction of a 'modern' (i.e. industrial)
society. This cliche is one of many to be disposed of in this
study. Still, this statement does point to the possibility of an
immensely productive relationship which could and should
be developed between contemporary European theory and
Japanese practice. And Marxism has always regarded such
mutually informative relationships between theory and
practice as essential to its growth.
One French author loosely associated with the Tel Que[
group, Roland Barthes, has in fact turned his attention to
Japan, and the essay which he produced following a brief
stay in that country, though a minor, circumstantial piece in
the context of his writings as a whole, is nonetheless a
pioneer text. 6 It is the first attempt by any Western writer to
read the Japanese 'text' in the light of contemporary semio-
tics, a reading informed by a rejection of ethnocentrism-
and indeed of all the 'centrisms' which have anchored
ideology in the West since the industrial revolution and the
rise of capitalism. I shall have occasion to refer again to this
essay, which has stimulated my own thought and research. 4· See Ueda Makoto, Literary and Art
The following, concluding passage of Barthes' book, which Theories in Japan.
contains the key to its title, points to the common ground of 5· Sieffert, Rene, La Tradition secrete
our concerns: du No, p. 56.
Empire of Signs? Indeed, if we imply that the signs are empty 6. Barthes, Roland, L' Empire des
and the ritual is a godless one. Just look at the study of signs signes.

I3
PREFACE (Mallarme's dwelling place), in other words, out there, any view,
be it urban, domestic or rural. And so that we may see how it is put
together, let it be illustrated by the Shikidai corridor: papered
with openings, framed by empty space and framing nothing, deco-
rated, it is true, but in such a way that the figuration (flowers,
trees, birds, animals) is swept away, sublimated, shifted far from
the forefront of vision ... in this corridor, as in the ideal Japanese
house, devoid or nearly so, of furniture, there is no place which in
any way designates property; no seat, no bed, no table provides a
point from which the body may constitute itself as subject (or
master) of a space. The very concept of centre is rejected (a
burning frustration for Western Man, everywhere provided with
his arm-chair and his bed, the owner of a domestic position).
Non-centred, this space is also reversible: you can turn the
Shikidai corridor upside down and nothing will happen other than
an inconsequential inversion of high and low, right and left. Con-
tent has been irrevocably dismissed: whether we pass through, or
sit on the floor (or ceiling, if you turn the picture around) there is
nothing to grasp. 7

All of the themes suggested in this dense text have their

place in my examination of Japanese film-making: the
essentially 'irreligious' character of the Japanese, their
rejection of anthropocentrism and of all the 'centrisms' that
derive from it in the West (the role which, in this respect,
their architecture has played in their films is absolutely
crucial, as Barthes indirectly suggests) and, of course, the
all-important 'irrevocable dismissal of content', i.e. of the
form-content hierarchies which are ours.
I have called my undertaking presumptuous, and some
explanations are in order. I do not read Japanese, and my
knowledge of the spoken language is elementary to say the
least. I was forced to rely upon interpreters in viewing the
vast majority of the films referred to here. I was. however,
able to study most of them at my leisure on an editing table.
As the reader will quickly perceive, my main concern is
with the modes of representation common to, and distinc-
tive of, most Japanese films within given periods, and with
the highly refined styles generated by these modes in the
work of a handful of auteurs (a concept here applicable only
with major reservations, as we shall see). The language
handicap is therefore not as serious as it might have been to
my 'content-oriented' predecessors in this field. More con-
stricting certainly are the objective conditions under which
the corpus of Japanese cinema is accessible to the scholar
today, and particularly, of course, to the visiting scholar. I
have spent altogether five months in Japan. I have had free
access to the four main collections of Japanese films: the
7· Barthes, op. cit., pp. 148-so. National Film Centre of the Tokyo Museum of Modern

14
Art, the Japan Film Council Library, the Kyoto Film Lib- PREFACE

rary and the private library of Matsuda Shunji. However, it

must be understood that less than three hundred complete
films made between 1897 and 1945 are preserved and read-
ily accessible in one of the above-mentioned archives. Sev-
eral hundred fragments collected by Matsuda Shunji are
also available for study; but that is all.
This does not mean, I hasten to add, that all of the
remaining tens of thousands of films produced by the
Japanese industry (from the outset, one of the world's most
prolific) are irremediably 'lost', despite the fact that there
was certainly a greater readiness in Japan to destroy com-
mercially obsolete films than in the West. 8 Since 1970, for
example, several early Ozu films, including the remarkable
Inn in Tokyo, all of which had been considered 'lost', were
unearthed in the warehouses of one of the major produc-
tion companies. To date, only about half of the forty-one or
forty-two films (see Appendix 2) directed by Ozu between
1927 and 1945 are known to be preserved. I am confident,
however, that more will be found as the number of Japan-
ese scholars and professional or amateur collectors gradu-
ally increases. 9 At present, their number is pitifully small,
considering the immensity of the task. It is, moreover, no
accident that the films brought to light in recent years have
been the work of acknowledged masters (such as Ozu and
Mizoguchi); in Japan as elsewhere, the rule of film conser-
vation dictates that discovery is determined by demand.
Ishida Tamizo, whose film Fallen Blossoms is analysed here
in some detail, is known to have directed scores of films; at
the time of my last visit to Japan, only two were known to
be preserved. The unanimous opinion of the small com-
munity of Japanese film scholars at that time was that, apart
from those two films, Ishida produced only 'pot-boilers'.
This had clearly deterred attempts to ferret out any other
films of his that might have survived and continued to
confirm, as well, his reputation as the maker of a single
'good' film. Or rather more accurately gave him no reputa-
tion at all, since even Fallen Blossoms, in my view one of
the masterpieces of Japanese cinema, is known only to a
few specialists - precisely because Ishida's name is not
attached to a body of work.
These cases are not exceptional; every pre-war director
constitutes a 'case', the most tragic being that of Mizoguchi,
8. For reasons which will become clear
whose mature work (1934-47) survives in only about half of as we go on to examine the Japanese
twenty films! Chance and arbitrariness necessarily, there- approach to the conceptions of
originality and authorship.
fore, inform these pages. I discovered Ishida quite by acci-
dent. None of the scholars consulted strongly recom- g. In illustration ofthis and in correction
of all the 1975 figures cited here
mended his films to me. Most of them were unfamiliar with regarding the conservation of films, see
his work. And while I have tried to see at least one film by Appendix 2.

15
PREFACE as many directors as possible from that golden age of the
Japanese cinema, the 1930s, I have had to follow, in large
part, a hit-or-miss approach. It is true that my more or less
random soundings did tend to confirm the existence of a
number of permanent traits which define a specifically
'Japanese approach', recognizable at every level of cultural
ambition. They also reinforced my view that Japan has had
only a few authentic 'masters'. Their work stands out in a
decisive way from the bulk of the national production,
either through the 'de-construction' of the Western codes
which, at certain periods in history, have tended to gain
prevalence there, or else through a supreme refinement
and systematization of those traits which are most specifi-
cally Japanese. 10 Of course, the random approach is always
in danger of neglecting a master or two; I might conceivably
have overlooked Ishida or, had I seen the wrong film first,
Shimizu Hiroshi. This arbitrariness should be underscored
here, if only because previous authors, whose work was
similarly conditioned, have failed to do so. It is not, how-
ever, incompatible with the purpose of a study which is, as I
have suggested, not a history of the Japanese cinema, but
rather a reading of a body of Japanese film in the light of
Japanese history.
If I wish to imagine a fictional people, I can make up a name for
them, treat them declaratively like an entity in a novel, found a
second Garabagne 11 to avoid compromising any real country in
my fantasy (but then it is the fantasy that will be compromised by
the signs of literature). I can also, without claiming to represent or
analyse any reality whatsoever (such being the major endeavours
of Western discourse) gather, somewhere in the world (out there),
a number of features (in both the pictorial and linguistic sense)
and with these features deliberately form a system. It is this system
which I shall call Japan. 12

Although the specialized nature of my present project pre-

cludes this radically theoretical stance, I do share the basic
premises of this opening statement of L' Empire des signes.
The reader may be struck by my reference to the 1930s
and early 1940s as the 'golden age' of Japanese cinema,
whereas most Japanese and Western specialists (the former
explicitly, the latter implicitly) situate it in the immediately
IO. Kinugasa in the I920S, Kurosawa in post-war period, the period of 'democratization'. The
the I950s are outstanding examples of reasons for this reversal will be analysed in their proper
the first approach, while Ozu and
Mizoguchi in the I 930s and early I 940s place. This period of Japanese cinema is terra incognita, not
perfectly embody the second. only to virtually all Western film-students but to most
1 I. A reference to Henri Michaux's
Japanese under fifty. It is my contention that the emphasis
mytho-poetic 'travelogue' Voyage en placed, in Europe, in America and, to a large extent, in
Grande Garabagne.
Japan, on, for example, the later films of Ozu and Mizo-
I2. Barthes, op. cit., p. 6. guchi, is due to the relative compatibility of those films with

16
the ideology of representation and signification which PREFACE
informs the dominant culture (and of course the dominant
cinema) of the West. The physical inaccessibility of earlier
but mature work by these directors can be seen, at least in
part, as a correlative of a fundamental incompatibility. For
if the films of that earlier period are unknown in the West-
Mizoguchi's Tale of Late Chrysanthemums and Tale of
Loyal Retainers in the Genroku Era are significant exam-
ples- it is largely because Western distribution, criticism
and cultural policy, as well as Japan's own scruples over
cultural commerce with the West, express the intuition that
such films are profoundly antithetical to canons of Western
cmema.
Needless to say, it is my hope, offered without undue
optimism, that the present study will stimulate interest in
the authentic golden age of Japanese cinema so that
hitherto unavailable films by masters, whether unknown,
like Ishida and Shimizu, or, like Ozu and Mizoguchi, cele-
brated within the Western cultural establishment as
assimilable to its own self-perpetuating values, can be seen
at last by viewers who no longer accept those values. As
Bertolt Brecht trenchantly put it: 'The only people who can
profitably study a piece of technique like Chinese acting's
A [lienation ]-effect are those who need such a technique
for quite definite social purposes.o~ 3 The people who most
need to study this cinema in its most 'radically Japanese'
form are those committed to constructing a thorough-going
critique of the dominant modes of Western cinema. This
critique, inscribed within seventy-five years of film practice
in Japan, remains unread. It is this reading which I hope to 13. Willett, John (trans.), Brecht on
initiate here. Theatre, p. 93·

17
 Some Terminological
Indications

The spatia-temporal audio-visual continuum which we call

a motion picture (or a video presentation or a slide-show
... ) may be regarded as a flux/field of signs. In the view of
Ferdinand Saussure - which has served to constitute a
semiology of the dominant cinema (that body of films sub-
ordinated essentially to the interests of the dominant class
and hence informed at every level by its ideology)- the sign
has two aspects, signifier and signified (Saussure compared
them with the two sides of a sheet of paper, indissociable
yet in binomial opposition). The signifier is the sign in its
materiality: it may be a pictorial element or set of elements,
or writing per se, a sound or set of sounds, or speech per se.
While this enumeration covers the substance of expression
peculiar to the 'audio-visual media', it by no means
accounts for all such substances. Setting aside the codes of
taste, touch and smell, rudimentary in our cultures, we may
say that all substances of expression are ultimately visual or
auditory, they may be transposed to film in either pictorial
or auditory representation (i.e. codes of clothing) or in
terms of 'form'- the codes of social distance translated into
shot sizes. This possibility of transposition provides one
foundation for the 'illusion of reality'.
The signified is the concept (and only the concept)
which, denotatively or connotatively, is ... signified. The
signified is not the referent. Even a word as 'abstract' as
philosophy, while it may signify a concept answerable to
some such definition as 'search for the underlying causes
and principles of reality', 1 also refers to an aspect of human
history, to an actual body of texts and practices. A referent
is 'that which refers a linguistic sign to extra-linguistic real-
ity as it has been articulated by a human group ... We
speak of a referential function when the message is centred
on the context.' 2 It is this concept of referential function
1. Webster's Third New International
which points to a provisionally useful, though incomplete
Dictionary. definition of the diegesis, helpful above all because it desig-
2. Dictionnaire de Linguistique
nates it as one dimension of film among others, not as its
Larousse, pp. 414-15. very substance. The diegesis may be said to be the imagi-

r8
nary referent or sum of referents. The density of the SOME TERMINOLOGICAL
diegesis, its presence, its powers of absorption, depend INDICATIONS
upon the status of the referential function. Most simply, the
diegesis is for instance the 'world of Balzac' which his
readers 'enter', those imagined drawing-rooms and cob-
bled streets, those imagined people, their 'characters', their
souls. (It should be clear that diegesis and narrative are in
no sense equivalent, since there is much more to diegesis
than narrative; the process of narration, which is part of the
narrative, is extra-diegetic.) In the cinema, of course, the
sign/referent relationship is clearly very different from
what it is in a written text, since we are dealing with an
audio-visual facsimile of the referent. Hence,.in film, the
referential function is always more heavily emphasized
than in literature.
The diegesis must not, however, be treated as a fixed,
simple object. It must be studied under two complementary
and dynamic aspects. For the constitution of the diegesis is
a process, and a more proper term would indeed be diegetic
process. It combines a mental process (the development of
the spectator's 'absorption'), and a process of 'writing': the
implementation on the screen of the 'codes' which catalyse
that absorption. Both processes are, to varying degrees,
implicated moreover in an ideological process. The resul-
tant is a diegetic effect, whereby spectators experience the
diegetic world as environment. Closely related and indeed
constituent effects have been designated in theatre as iden-
tification (Brecht) and in painting as illusionism. However,
it is also essential to consider the place of the audio-visual
diegesis in the historical process. The stages of formation
and the on-going elaboration of the basic mode of rep-
resentation in the cinema of the West point to an all but
universal tendency within the dominant cinema (as well as
theatre and literature) to maximize and generalize the
diegetic effect. And this tendency, in turn, must be under-
stood as entering into a dialectical relationship with cul-
tural and ideological conditioning.
This maximization of the diegetic effect implies the effac-
ing of the sign and of the process of signification as con-
sciously perceived, and is only incidentally related to
ideological categories such as 'realism' or 'naturalism'. The
diegetic effect is truly as 'strong', as unadulterated in Hell-
zapoppin, The Five Thousand Fingers of Doctor Tor late
Fellini as it is in Ladri di Biciclette. Conversely, it is still
intermittently operative for many spectators of Michael
Snow's Wavelength or Jean-Luc Godard's Deux au trois
chases que je sa is d' elle. Here, it is true, the status of the
diegesis changes, qualitatively and quantitatively, from one
moment to the next. These films designate the diegetic
SOME TERMINOLOGICAL effect as such, making it difficult or impossible for the
INDICATIONS reader-spectator to enter 'permanently' into the imaginary
space-time constituted through the dicgetic process. Films
of this sort function as critical texts.
As indicated in Chapter 5 the development of the 'lan-
guage' of the dominant cinema was principally determined
by the need to improve the effectiveness of diegetic produc-
tion, to intensify the presence of characters and 'their'
world. The so-called Griffith codes of editing, and con-
comitantly of framing, constitute one of the three main
supports of that production. To it were added the norms
of 'three-dimensional' lighting/ composition, developed
chiefly during the 1920s, and, of course, lip-synch sound. To
what extent are we justified in calling these basic systems,
which obviously correspond to such different orders of
semiosis, codes? A code may be defined as 'a system of
signals - or signs or symbols, which by pre-established
convention, is intended to represent or to transmit informa-
tion between the origin or transmitter of the signals and the
destination or receiver.' 3 The notion of pre-established
convention has been so generally extended by the science
of semiotics to cultural and ideological determinations of
an 'unconscious' nature (e.g. the codes of clothing) that it is
reasonable to postulate the existence of codes of expecta-
tion with regard to the established procedures of editing,
for example. True, the information transmitted by the
dominant 'codes' of editing is elementary and most often
negative (e.g. 'no time has elapsed between this shot and
the last'). It is also true, however, that their structural
functioning is in accordance with the considerably more
complex codes of, say, any given genre, which will combine
certain specialized and sophisticated codifications of the
'ground-rules' (for example, shot-size codes) with signify-
ing patterns of a 'purely diegetic' character (dialogue, set
design, etc.). 4 The 'editing codes' arc in fact part of the
foundation supporting the more complex codes of the
dominant system. When this is understood, the semanti-
cally elementary principles of dominant editing (together
with the norms of frame-composition and lighting, and the
technology of synch-sound) are seen to constitute a mode of
representation which is primary with respect to the codes of
genre, among others. Western film has known, prior to
around 1912, a period dominated by a different mode of
representation, the 'primitive mode', which has had a sig-
nificant counterpart in Japan ever since the introduction of
3· ibid, p. 92.
the Western invention. This claim, essential to our inves-
4· When these 'diegetic' codes depend tigation, will be substantiated by comparative analysis.
so1ely upon visual representation of
socially pre-existent systems, they are
The concept of matching is crucial to the dominant sys-
called iconographic codes. tem of editing/camera placement. Practice has come to

20
define the match as any element which guarantees, from SOME TERMINOLOGICAL
one shot to the next, the illusion of diegetic, spatio- INDICATIONS
temporal continuity- an 'illusion' distinct from the diegetic
effect per se, though constitutive of it. Thus, the fact that a
character is holding the same high-ball glass filled with
what appears to be the same amount of liquor from one
shot to the next constitutes a correct match. If, on the
contrary, the character appears, in the second shot, to have
lost the glass with an abruptness defying verisimilitude or to
be nursing noticeably more or less whisky than before,
without having drunk or been served- or been hidden from
view long enough for one or the other to have occurred- we
are dealing with a mismatch, a bad or incorrect match.
More fundamental, however, is the set of orientation
match-cuts whose development between 1900 and 1930
may be said to provide the framework of the history of the
dominant cinema during the silent and early sound eras.
This development appears, in part at least, to have been the
material consequence of the growing awareness among
film-makers that perceptual left-right screen-orientation
had to be given priority over actual orientations to pro-
filmic space, the space of the 'studio'. Two characters look-
ing at each other in pro-filmic space may or may not appear
to the average Western spectator to be looking at each
other in two shots appearing successively on the screen,
accordingly as the respective placings of the camera did or
did not comply with certain conditions. As was generally
agreed, although only after some twenty years of film-
history, what matters is that the eyelines, the orientation of
gaze with respect to lens/screen-centre, must appear to
'meet', to cross in an imaginary space formed by the placing
side by side (and not 'face to face') of the two successive
shots. Similarly, a character walking in a straight line along
a road through pro-filmic space who exits screen-left in one
shot and re-enters screen-left in the following shot (because
he has, for example, been filmed from the opposite side of
the road) may well appear, in contradiction with all other
data supplied, to have abruptly retraced his steps. This
anomaly is termed an incorrect (screen-) direction match.
Similarly, reversals of the respective positions of people or
objects on the screen can, under certain conditions, be
equally disruptive of diegetic verisimilitude. As evidenced
in the example of the apparent reversal of walking, the I8oa
match (not to be confused with the reverse-field) often
brings about reversals of all three types, which are in fact
merely different manifestations of the self-same phenome-
non.
The eyeline match and to some extent the direction
match ensure temporal continuity and spatial communica-

21
SOME TERMINOLOGICAL tion between shots designated as contiguous in diegetic
INDICATIONS space. The most common device used to establish and to
exploit an impression of contiguity/continuity is a series of
eyeline exchanges which I have chosen to call reverse-field
set-up or succession of reverse-angle shots. 5 A reverse-field
series may occasionally involve r8o 0 reversals; usually,
however, these will produce 'bad' matches.
A contiguity cut is one which implies spatial contiguity
and temporal continuity. The reverse field is only one,
albeit the most important form of the contiguity cut. 6 In
terms of spatial (as opposed to temporal) relationship,
there are two other basic modes of shot-change: the direct
cut or direct match, in which two successive shots contain
common visual elements, present two successive aspects of
a single fragment of diegetic space; and the alterity cut, a
shot-change which involves neither spatial nor temporal
continuity at all in the ordinary sense, but may present a
considerable range of temporal or other semantic varia-
tions on the alterity mode. Historically, it was the first type
of shot-change, the purely disjunctive equivalent of the
curtain between scenes on the theatrical stage, which
derived spatia-temporal signification only from the
appropriate indications in the text of the play (or the pro-
gramme). The alterity cut was also the first shot-change to be
clearly and massively encoded: as ellipsis ('time has elapsed
between the end of the last shot and the beginning of this
one'), and as ( unsituated) cutaway ('this shot is taking place
"elsewhere" with respect to the previous one'). 7 The
dialectical development of these two functions produced
the simultaneous mode of cross-cutting ('the separate
actions shown in these alternating shots are taking place at
the same time'). The era-to-era cross-cutting at the end of
Intolerance, the rhetorical associations of Strike, October,
The Man with a Movie Camera, etc. are advanced
5· The procedure is sometimes called developments of the alterity cut. They create far more
'shot-counter-shot' in excessive fidelity
to the very common French term
complex problems of definition, which the classical distinc-
champ-contre-champ. However, the tion between parallel montage and alternate montage is
word 'reverse' docs figure in a number
of related expressions employed among
quite inadequate to resolve.
British and American professionals. The earliest form of direct cut (c. 1905) was the concer-
6. The cutaway may also operate on the
tina, i.e. a 'cut-in' or 'cut-back' to or from a given subject
contiguous mode, which may in fact be which remains the centre of each successive shot composi-
shown to have derived historicallv from
that of alterity. ·
tion.8 This was to be one of the chief centering strategies of
the cinema after Griffith. The 90° match and, ultimately.
7· Sec my Theorv of Film Practice. ch. 1.
the r8oo reversal were introduced gradually over the next
8. The term concertina seems no longer decade and a half, and the development of the contiguity
in current usage in the British film
industry where it originated. However. I
match and reverse field, initiated somewhat later, took a
find it the most satisfactory equivalent comparable period of time (191G-25). Matching on move-
for the common French term raccord
dans l' axe provided by actual practice in
ment developed somewhat more slowly (1905-25), in reac-
English. tion to the disconcerting discovery that the direct cut was

22
not necessarily the least disruptive of the three modes once SOME TERMINOLOGICAL
the other two had been properly mastered. The direct cut INDICATIONS
proved in fact to be more ofa 'jolt' than the other forms
unless the kinetic effect of the cut were absorbed by a
shot-to-shot chain of perceptually continuous diegetic
movement.
Finally, though I refer often to these 'rules' and practices,
and to the principles that subtend them, as 'editing codes',
they are of course, strictly speaking, the codes of decoupage
(i.e. of the editing/camera set-up relationship). The success
of orientational matching is ultimately contingent upon the
'correct' placing of the camera, while the very possibility of
matching on movement depends upon the placing of cam-
era and actor(s).

23
Partl Grounds, Premises
1. A System of Contradictions

A great deal has been written about 'Japanese culture'

since the opening of the country to foreign exchange, much
of it uninformed nonsense from the pens of well-meaning
globe-trotters, sensitive indeed to the otherness of the East
but able to comprehend it only as the last refuge of 'univer-
sal human values' - those of middle-class idealism. One
feels, therefore, reluctant to enter into general discussion
of Japanese modes of representation, of Japanese aesthe-
tics, of the 'Japanese mind', reluctant to add, perhaps, to
that century-old slag-heap. It is nevertheless impossible to
deal with Japanese film, indeed with any aspect of Japanese
society, without constant reference to almost every one of
its other aspects. That this should be especially true of
Japanese society is an assertion which in itself requires
some historical substantiation.
The pertinent traits of Japanese aesthetics were defined
almost entirely between the ninth and twelfth centuries,
known as the Heian period.lt was, however, under the rule
of the Tokugawa Shoguns ( r633- r 867), the first of whom,
Ieyasu, is celebrated as the consolidator of national unity
and the initiator of three hundred years of peace, that
Japan became the mostintegrated large-scale society in the
world, East or West.

Although the Shogunate thought to preserve its power by main-

taining carefully drawn class distinctions, and although the end of
the Tokugawa period was a time of social ferment, the long
isolation of the country created a uniformity of habits, beliefs, and
tastes among all classes. The farmer, the townsman, and the
samurai moved in different hierarchical spheres but they shared a
mundane world in which their differences were not of kind but of
degree. The diet of all, except in times of famine, was essentially
the same, consisting principally of rice with a few vegetables and a
little fish, prepared according to a very few standard recipes ....
The house of a townsman was more elegant than that of the
peasant, but the architectural arrangement, the household arti-
cles, the system of heating, the furniture -all had the same basic
GROUNDS, PREMISES form, size, shape and function. The wealthy townsman could
warm his hands in winter over a charcoal brazier made of bronze
while the peasant had to be content with one of clay and with
somewhat less charcoal ... Wealth could provide greater abun-
dance of food and better quality materials; it could purchase
nothing essentially different in form or function. The code of
morality and social behavior, based largely upon the Japanese
interpretations of the commentaries of Chu Hsi on the works of
the Chinese sages, at last permeated into all social groups, so that
finally the Edo fireman shared an ethical climate not immediately
distinguishable from that of the samurai.
Similarly, tastes in art were built up out of commonly shared
artistic experiences .... Identical tastes in music and in the
graphic arts were so widespread that, until modern times, the
body of artistic canon did not differ significantly according to
social stratification. A maid serving tea moved in a style similar to
that used at court ceremonies; shop clerks were taught to sit,
move, and stand in patterns used in the No and were intensively
drilled to speak in a manner closely resembling that of the Kabuki
actor.'

Even the Japan of the 1970s, after a hundred years of

capitalism and a thirty-year inoculation of American indi-
vidualist ideology, still bears the unmistakable stamp of
three centuries of standardization (e.g. the ken or module
upon which tatami matting, and all other household furn-
ishings, as well as all traditional dwellings, are patterned).
And in the first forty years of this century, the period during
which Japan's principal modes of filmic representation
developed, the above description of Tokugawa society was
still remarkably relevant. Small wonder, then, that so many
of the specific traits of Japanese society are explicable and
comprehensible, not as 'borrowings' from the older arts but
rather as specific manifestations of that vast circulation of
modules -of 'empty' signs, as Barthes might say- which is
the very substance of Japanese society.
Before attempting to define some of the principles go~­
crning this circulation, it may be useful to respond to one
question which arises at the inception of any historically-
oriented investigation of Japanese cinema: how is it that
while several of the non- Western nations - China, of
course, but even more prolifically, Egypt and India- have
long produced motion pictures, only Japan has developed
modes of filmic representation that are wholly and specifi-
1. Earle, The Kabuki Theatre, pp. 67-ll.
cally her own, only Japan has produced a body of 'master-
2. The inverted commas here call
attention to a dubious but for the
pieces'.2lndeed, an observation made by Jay Leyda about
moment unavoidable reference to the the transition from silence to sound, in his very informative
occidental concept of the unique,
irreplaceable work, a concept irrelevant
history of the Chinese cinema, seems to characterize the
in the traditional Japanese context. entire history of cinema as it unfolded, not only in China

26
but in India and Egypt as well: 'The limitations of the A SYSTEM OF
American sound film were taken over wholesale by Chin- CONTRADICTIONS
ese producers when they finally achieved sound produc-
tions of their own, and it cannot be said that Chinese films
have yet escaped from their entangling admiration of the
efficient American film.' 3 Leyda shows how the entire
Chinese industry depended throughout its formative years
(basically, up to 1930) on American and English capital
and, most important, perhaps, on American and English
cameramen. Yet China was not even a 'full-fledged' colony
(though much actual production does seem to have been
concentrated in Shanghai and Hong Kong, both dominated
by European interests). To what lengths, one wonders, was
the reliance on 'Western know-how' carried in 'protecto-
rates' or colonies of such long-standing as Egypt and India
during the formative period? In the absence of any com-
parably authoritative histories of the Egyptian or Indian
cinemas in French or English, I am reduced to conjecture.
Both nations remained under British rule until after the
Second World War. Since the British cinema has from its
very inception been totally enthralled by what were to
become the 'Hollywood modes and codes', it is not difficult
to understand why the modes of spatia-temporal represen-
tation found in the films of those countries appear to be so
utterly divorced from the representational systems at work
in their traditional arts.
Even a cursory examination of Asian history immedi-
ately provides the answer we are looking for: in two
thousand years of recorded history, no part of Japanese
territory had ever been occupied until the 1945 defeat.
Japan was never subjected to the semi-colonial status
which was China's for over a century, or to complete
enslavement as were Egypt or India. She is the only major
non-Western country to have escaped the colonial yoke.
Her cinema, of course, is but a minor consequence of this
crucial fact, which in turn is one of a series of historical
features by which Japan may be identified as belonging
simultaneously to several apparently incompatible types
and stages of historical development. The definition of
these primary traits, essential for the understanding of the
secondary traits as evinced by Japanese films, is our next
task.
The fact that Japan was able to avoid colonial rule in the
nineteenth century (by 'accepting Western influence', i.e.
using Western technology to build a bulwark against West-
ern imperialism) contributed directly to the originality of
the Japanese film, since it made possible the technical and
economic autonomy of her film industry. From a very early
date, Japan trained her own technicians, developed and 3· Leyda, Jay, Dianying, p. 64.

27
GROUNDS, PREMISES printed her films in laboratories owned and operated by
Japanese; she even manufactured her own raw film-stock. 4
And, of course, this self-sufficiency of the infrastructure was
a sine qua non for a 'free' interaction between the
cinematograph and the cultural milieu.
Historically speaking, Japan's avoidance of the 'colonial
stage', the corollary of the construction of a modern capital-
ist state, carried out at twice the pace of advanced Western
nations, was determined in part by a prior condition,
another unique feature of Japanese history. Although
Japan's 'Asiatic-ness' is undeniable (geographical situa-
tion; agrarian, rice-dominated economy), she is also the
only major Asian nation never to have known what Marx
defined as the Asiatic Mode of Production, characterized
from earliest times in China, India, Persia, Egypt, etc., by
vast hydraulic enterprises entailing the mobilization of
thousands of labourers and consequently the early estab-
lishment of a powerful central government with a large civil
service in its employ, and a drastic curtailment of city
autonomy. This form of government, analysed by K. A.
Wittfogel in his classic study of the hydraulic society, 5 has
been termed 'oriental despotism'. Marx had observed,
however, that 'Japan, with its purely feudal organization of
landed property and its developed petite culture, gives a
much truer picture of the European middle ages than all
our history books, dictated as these are by bourgeois pre-
judices.'6 Wittfogel sheds further light on this anomaly:
4. This was in keeping with the general
policy of early Japanese capitalism: Why did Japan's rice economy not depend on large and
'Foreign instructors, advisers and
engineers were brought in to run a
government-directed waterworks? Any competent economic
number of the new factories and train geographer can answer this question. The peculiarities of the
the technicians who were to run them in
the future, as many as 130 being
country's water supply neither necessitated nor favored substan-
employed by the Department of Public tial government-directed works. Innumerable mountain ranges
Works alone by 1879· Official policy.
however, was to replace them as soon as
compartmentalized the great Far Eastern islands; and their
possible by Japanese whose salaries broken relief encouraged a fragmented (hydroagricultural) rather
were smaller, an attitude which led one
British resident to observe that ''the
than a coordinated (hydraulic) pattern of irrigation farming and
Japanese only look upon foreigners as flood control ...
school-masters. As long as they cannot
help themselves they make use of them:
The rulers of the dominant political center effected a loose
and then they send them about their political unification at a rather early date, but they were not faced
business." It was precisely this. of
course, that eventually made Japan's
with hydraulic tasks that required the coordinated operation of
industrial technology self-sustaining, in large corvee teams. Nor were they conquered by the forces of an
contrast to that of other Asian countries,
which remained for the most part
orientally despotic state. They therefore failed to establish a com-
dependent on foreign help'. (Beasley, W. prehensive managerial and acquisitive bureaucracy capable of
G., The Modern History of Japan, p.
144. The interpolated quote is from 0.
controlling the non-governmental forces of society as did the men
R. Black's Young Japan, I88o-I; the of the apparatus on the Chinese mainland ... Many elements of
italics are mine).
Chinese culture notwithstanding, the decentralized and pro-
5· Wittfogel, Karl A., Oriental perty-based society of the Japanese Middle Ages resembled much
Despotism.
more closely the feudal order of the remote European world than
6. Marx, Karl, Capital, p. 718. the hydraulic patterns of nearby China.
However, Wittfogel goes on to add significantly: A SYSTEM OF
CONTRADICTIONS
We must be careful not to oversimplify the picture. The Oriental
quality of many Japanese institutions and ideas is beyond doubt.
On the lower and local level, Japanese irrigation agriculture
required quasi-hydraulic coordination and subordination; and
feudal lords' insistence upon absolute obedience may, at least in
part, reflect such quasi-hydraulic relations. Rudiments of a postal
system seem to have existed prior to the Tokugawa period; ...
The members of the ruling group, although strongly imbued with
a military spirit, continued to think in terms of somewhat adjusted
Confucianism; and although they invented simplified phonetic
symbols, they employed with genuine pride the Chinese script,
which, like Confucius' conception of the gentleman-bureaucrat,
was better suited to a civil and learned officialdom than to a
war-minded knighthood. 7

(We shall explore later and at some length the crucial

matter of the Japanese writing system, an essential emblem
of the multiform bivalence implicit in this description.)
Japan was related 'in a submarginal way' as Wittfogel
puts it, 'to the institutional patterns of the Asiatic world'
and this undeniably made the islands a privileged terrain
for the extra-mural development of important aspects of
Chinese culture. Nevertheless, the remarkable parallels
between Japanese and European feudalism undoubtedly
made possible the rapid conversion during the Meiji period
(r867-1912) to a unique variety of state capitalism, involv-
ing an immediate priority to heavy industry, a conversion
paradoxically contingent upon the perpetuation of feudal
social structures. 8 For Japan is also the only major capitalist
nation in the world today which has never known a true 7· Wittfogel, Karl A., Oriental
Despotism, pp. 197-200.
bourgeois revolution in any guise whatsoever. 9
The Meiji Restoration of r867 in no sense constituted a 8. See Norman, E. Herbert, Japan's
Emergence as a Modern State.
sudden shift of power from the feudal classes to the
bourgeoisie. Insofar as this shift took place at all, it did so 9· Bismarck's construction of German
industrial capitalism 'from the top
over a period of almost a century and hence so gradually down' bears some relation to the similar
that the two classes in many ways may be said to have stage in Japan, and it is no accident that
those who drafted the Meiji constitution
interpenetrated (as to some extent they already had under ( 1889) took as their model the
the Tokugawas); which of course goes a long way towards constitution of Germany. However,
fundamental differences subsist
explaining the longevity of ancient customs and modes of between the histories of the two
thinking. This fact is essential to any attempt to understand countries, as witness in particular the
fact that while Nazism was the solution
twentieth-century 'bourgeois ideology' in Japan, where the to the crisis of capitalism adopted by a
term is perhaps not even adequate. As for the events of German middle-class which, after the
failure of the proletarian revolution,
r867, the Meiji Restoration was simply the linchpin of indisputably held the reality of power,
Japan's uniquely concatenated transition to capitalism. As the Japanese middle class was still an
opposition, 'liberal' force during the
Sir George Sansom points out: period of exacerbated militarism which
preceded the Second World War and
The driving force at the Restoration and in political life for the which constituted the neo-feudal
aristocracy's final attempt to perpetuate
best part of a generation thereafter was provided by leaders who its absolutist rule.

29
GROUNDS, PREMISES had been brought up in a feudal, or at any rate feudalistic, atmos-
phere. They were for the most part dissatisfied and ambitious
samurai, and their outlook was coloured by their antecedents. It
was these men. and not the bourgeoisie, who laid the foundations
of a capitalist structure and at the same time developed a political
system that bore little resemblance to those which came into force
in the advanced industrial countries of Western Europe under the
influence of a powerful moneyed class. 1"

As for the agrarian aspects of the Restoration, Takahashi

Kohachiro sums up thus the conclusions of Japanese Marx-
ist scholars:
The agrarian reform of the Meiji Restoration made the jinushis
the actual owners of the land, but they had nothing in common
with the English landlords who let their land to farmers who, in
keeping with the capitalist mode of production, were in turn
agricultural entrepreneurs employing wage-earners. Nor do the
jinushi have anything in common with the squireens of Eastern
Europe who farmed their own land directly with serf-labour ...
The jinushi divided his land into small lots and had it worked by
small farmers (kosakus) who paid as rent nearly half of their
harvest, like the poverty-stricken melayers under the old regime in
France ... Thus historical comparison and perspective provide
proof that the Meiji Restoration was not, properly speaking, a
bourgeois revolution of the 'classical' type. 11

A prosperous non-European nation which was never col-

onized, an Asian nation with only a 'submarginal' experi-
ence of the Asiatic Mode of Production, a nation which
moved from feudalism to full-fledged capitalism without
the hiatus of revolution: there are anomalies here and it is
interesting to consider what role they have played in
determining the constitution of a specifically and uniquely
Japanese 'outlook', and what their bearing is on that
'Empire of (empty) signs' described by Roland Barthes. A
comprehensive answer to such a far-reaching question is
not only beyond the scope of this book but beyond the
capacities of the avowedly amateur Japanologist that I am.
I can only point to the direction in which I feel the ans-
ro. Sansom, George B., The Westem wer(s) may lie, and occasionally, with more special refer-
World and Japan, p. 327. W. G. Beasley
(op. cit., pp. 224-5) demonstrates that ence to 'film history', contribute observations which I feel
the domination by the oligarchy issued relevant to the problem.
from this class remained complete until
the end of the Pacific War. The organs Now it is at this point in our undertaking that we un-
of expression of the new middle class as avoidably encounter another pair of 'contradictory' traits,
such- the Parties and Diet- had no
more real power than the Tiers Eta! exhausted by specialists of 'things Japanese'. They are
under Louis XVI. described, depending upon the writer's inclinations, either
I 1. Takahashi Kohachiro, 'La as 'faculty for assimilation' and 'sense of tradition' or,
Restauration de Meiji au Japon et la alternatively, as 'lack of originality' and 'stagnant conser-
Revolution Fran<;aise', in Recherches
lnternationales, no. 62, 1970, pp. 84-s. vatism'. In either case, they are seen as sure symptoms of

30
the 'irreconcilable dichotomy' of the 'unfathomable' A SYSTEM OF
Japanese character. We must of course move beyond these CONTRADICTIONS
stereotypes, but we cannot bypass them: an analysis of the
ideologies which they conceal is crucial to the present
undertaking. Nothing has better served detractors of
'Japanese culture', nor more keenly embarrassed the bulk
of its Western admirers, than the ease, indeed the eager-
ness, with which the Japanese, at certain periods in their
history, have adopted techniques and concepts of foreign
origin - among which the cinema and the modes of rep-
resentation attached to it in the Western context are not the
least insignificant. In drawing the inevitable comparisons
between Japan and China, those who recently saw China as
an ally in the struggle against fascism, those who see her
today as the vanguard of socialism, were/are quick to com-
pare the 'originality' of China's 'age-old wisdom' with
Japan's 'habit of copying'. Were not Japan's writing and
much of her vocabulary, broad sectors of her arts and her
most 'sophisticated' religion 'lifted' wholesale from China?
The ease with which she has adopted Western techniques in
modern times is merely additional proof that her only
'aptitude' is a mimetic one. The champions of Japanese
culture- whose number has multiplied rapidly, especially
in the United States, since Japan has become 'a bulwark
against Communism' -are quick to counter, but they too
invoke the value of originality: the Japanese do not copy,
they adapt; whatever has been borrowed has been pro-
foundly and creatively transformed, and besides, many
aspects of Japanese culture are entirely indigenous (and the
list adduced in evidence is a long one, from kabuki theatre
to a dish like sushi and from the Tale of Genji to origami
paper-folding. Of course, there is a good deal more 'truth'
in the second type of statement than in the first; what
interests us here, however, is that both arguments have an
ideological base since both call upon the virtues of origina-
lity. And this is where Japan has the greatest power to
perturb, for in Japan, as I will abundantly illustrate, and
notably in her cinema, originality has never been a domi-
nant value. In particular, the specifically bourgeois notion
that the artist is the creator and proprietor of his work is
utterly meaningless within the framework ofthe traditional
arts of Japan, and of the East in general. Plagiarism seems
to have been a meaningless concept until the opening of the 12. That this notion is intimately linked
with the growth of capitalism is amply
country. Here, then, is a fundamental point of conflict with demonstrated by the acknowledged
the modern West, where the notion of the artist as intertextual practices that continued
right up to the French Revolution
demiurge, creator of works which are his inalienable prop- (Chaucer's and Shakespeare's extensive
erty, has been an essential article of faith since the end of borrowing of forms and material are
celebrated; Michel Corrette's choral
the eighteenth centuryY The Japanese social system adaptation of a Vivaldi concerto is an
denies the very concept of originality, acknowledges and exemplary illustration).

31
GROUNDS, PREMISES indeed deliberately emphasizes the material reality of the
circulation of signs. This system erodes the very founda-
tions of our ideology of the Creator as the Supremely Free
Human Being, of the Artist-God, of the Book and the
Word.
The Japanese attitude here described is bound up with
modes of thinking that go back to very ancient times (we
shall see some of its earliest manifestations in Heian liter-
ary practice), and this preservation is the other term of our
hackneyed dichotomy, traditionalism. It is true, of course,
that the Japanese as a nation have always tended to 'keep
things': once it's in the house, once it's proved its useful-
ness, you never throw it away. Trivial as this formulation is,
I believe it adequately describes a phenomenon epitomized
by the Imperial House with its 'two thousand years' of
unbroken continuity, and perhaps even more vividly by the
Shosoin at Nara, 'a warehouse of all the useful objects of
the court given by the Emperor Shomu in AD 756 and
almost completely preserved to the present day in its origi-
nal log storage building'Y The survival of such massive
fragments of 'tradition' is largely explained by the above-
mentioned lack of any revolutionary disruption in Japanese
history, by the gradualness of the elimination of feudal
institutions, and by Japan's long period of isolation. This
'sense of tradition' is most remarkable in the permanence
of its forms. Once these achieve stability, they tend to
remain 'as is' for centuries. Of this complex and strange
phenomenon, even the casual visitor to Japan is immedi-
ately aware.

In the Meiji period, Japanese with modern tastes built into their
homes a Western Room, furnished in the late nineteenth-century
style of interior decoration, and such rooms, although they have
largely ceased to exist in the West except in museums, remain a
necessary adjunct of the well-appointed Japanese home. Some of
these rooms are today decorated in a style that can be described
loosely as Swedish Modern, but the majority repeat the doily, the
antimacassar, the fringed table cover, the elaborate gilt picture
frame, the floral carpeting and the busy design of chair and table
which characterized the Victorian parlor. 14

Many Western ideas have also been poured into a similarly

permanent mould: by and large, contemporary 'liberal' and
even 'socialist' thinking in Japan bears a remarkable
13. Lee, Sherman E., A History of Far
Eastern Art, p. 272. resemblance to the ideas gleaned from the writings of
Spencer, Mill and Rousseau as they were propagated in the
14. Ernst, Earle, op. cit., p. 272.
r88os by Tokutomi Soho, the intellectual leader of the
15. See Pyle, Kenneth B., The New reformist opposition to the Meiji government. 15
Generation in Meiji Japan. However,
this thesis is entirely my own. Why is Japanese society the most 'conservative' in the

32
modern industrial world? It is true, of course, that for three A SYSTEM OF
hundred years the Tokugawa Shogunate went to extremes CONTRADICTIONS
to preserve the status quo, encouraging infanticide in order
to maintain the population level, allowing bridges and
roads to fall into disrepair so as to discourage trade, etc.;
conservatism and its 'fixative effect' can, however, be
traced back another thousand years, to Japan's earliest
history. It is no doubt related to the extremely difficult
natural conditions under which the Japanese nation was
built, in a tiny archipelago of which only 16 per cent of the
land is arable, ravaged by typhoons, torrential rains, land-
slides, earthquakes, tidal waves and volcanic eruptions. In
such precarious circumstances- and protection against the
elements remained problematic until well into the twen-
tieth century - a predominantly agrarian population will
quickly seize upon any new tool which may seem useful, at
the same time refusing to discard the old, as it is associated
with so many tragic struggles, and may well serve again if
the new tool fails. Whatever its origins - and they are
certainly more complex than I can suggest here -the 'fixa-
tive effect' has played an absolutely vital role in the history
of Japanese culture and particularly in the development of
the arts including, as we shall see, the cinema. Its historical
patterns are aptly described by the critic, Kato Shiiichi:
... different types of art, generated in different periods, did not
supplant each other, but co-existed and remained more or less
creative from the time of their first appearance up to our time.
Buddhist statues, a major genre of artistic expression in the period
from the sixth to the ninth century, continued to evolve in style
during the following eras, even when the picture scroll opened
new possibilities for the visual representation of the world in the
Heian period. Brush works with india ink flourished during the
Muromachi period [ 1336-1573], but one school of artists
remained faithful to the techniques and style of the picture scroll.
Under the Tokugawa regime, a new style of painting and decora-
tive art was established by Sotatsu and Karin; the technique of the
woodcut print was elaborated to perfection. Yet artists never
ceased to carve Buddhist statues or engage with great passion in
brush-work painting. Practically no style ever died. In other
words, the history of Japanese art is not one of succession but one
of superposition. 16
The contrast, we might add, with Western art, is absolute.
We encounter once again the question of originality, for the
birth and death of forms and styles in the West have long
been regulated by the law of supersession (and of self-
consciously neo-classical revivals) and this can be observed
as easily on the walls of a museum as in the pages of a 16. Kato Shiiichi, Form, Style and
history of fashion. Conversely, the principle of superposition Tradition, p. 4·

33
GROUNDS, PREMISES described by Kato may be readily observed in the streets
of any Japanese city today, where Western-style business
suits and blue jeans co-exist unostentatiously with clothes
that have not changed in any significant detail for over
three hundred years, which still occupy an entire floor of
any popular department store. And, as we shall see, the
cinema of Japan has borne at almost every stage the stamp
of this superposition, even up to the present day.

34
2. ASystem of Signs

It is in the period of the Fujiwara Regents 1 that there were formed

among upper-class men and women in privileged situations cer-
tain standards of personal behaviour and certain canons of aesthe-
tic judgement which are the source, or it might be better to say the
foundation of Japanese social life as it developed in later cen-
turies.2
It follows from this judgement by the West's most eminent
historian of Japan that a detour through the social and
literary practices of the Heian court is indispensable if we
are to bring to light the theoretical implications of Japanese
aesthetic practice as a whole. And our first task will be a
cursory consideration of the writing system which made
Heian literature materially possible and which in itself
contained some of the seeds from which the forms and
styles of that period, its writing in the strongest sense, may
be seen to grow.
We know that it was long before the Heian period,
probably in AD 552, that Chinese writing was first intro-
duced to Japan by way of Korea, in several volumes of
Buddhist scriptures, which were also to initiate the spread
of the great continental religion to the island empire.
Almost from the outset it seems that while the learned
clergy strove to master the Chinese language as a written
(and occasionally spoken) whole, efforts were also made to
adapt the Chinese characters to the forms of spoken Japan-
ese, a polysyllabic, inflected language, quite in contrast on
these two fundamental counts to Chinese.
The methods used were several and complex, but in 1. The reign of the Fujiwara family as
summary we may say that a partial phoneticization of the the power behind the throne is
associated with the first of several
essentially non-phonetic Chinese writing was undertaken: periods during which the imperial
the sounds of spoken Japanese were assigned to characters capital, Heian-Ky6 (later Kyoto), was
the seat of actual political power
whose pronunciation in Chinese was thought to approxi- (794-II91).
mate those sounds, while at the same time these same
2.Sansom, George, A History ofJapan,
characters - and many more whose original sound may Vol. 2, p. 73·
have been irrelevant to the native language - were also
3. The main source here is Sansom,
used to represent 'ideographically' Japanese words of George, An Historical Grammar of
equivalent meaning. 3 Japanese.

35
GROUNDS, PREMISES This method of writing (kana majiri, i.e. mixed phonetic
writing) made possible the first written poetry, recorded iu
the early chronicles and in the first of the great imperial
anthologies, the Manyoshu (759). However, this phonetic
use of the Chinese characters was cumbersome at best: the
ambiguities involved in the double function of certain sign~
made reading an arduous task, while the inflexions of
Japanese verbs and adjectives often required as many as
five or six complex characters to express phonetically what
in pure Chinese would have been a single sign/concept
Gradually, however, during the course of the He ian period,
by a process of stylization and codification, a specifically
Japanese syllabary (hiragana) was developed, partly in
relation to a general reaction to the overwhelming ChinesL
influence of the previous three centuries, partly in responsL
to the growing needs of both the administration and coUJ
society. Not only did the new script 'liberate' the ChinesL
characters (kanji) from the burden of phonetic expression
but it promoted the emergence of one of the world's great
literatures. For a long time, kana majiri in its many forms
continued to be assiduously practised by statesmen auJ
clergy in their non-literary writing, while the great novels
and poetry of the Heian period were written in a script
largely dominated by the phonetic system. The ultimate
result of this co-existence of two fundamentally different
types of writing, the one phonetic and the other primarily
non-phonetic, was the creation of a unique system which
incorporates both. 4
The walls of a Tokyo subway station today bear signs on
which the name of the station can be read in three different
scripts: kanji, since every Japanese proper name possesses
a Chinese character or set of characters which has become
attached to the phonetic signifier through a process of
historical selection; hiragana, since of course the syllabary
may at least theoretically be used to transcribe any word
used in the Japanese language, even if it is not normally so
written (or to render, in terms of Japanese pronunciation,
any words of foreign origin that need to be incorporated
4· The Korean writing system has, tor into the language); and finally rvmaji or Roman characters.
three hundred years, been similarly
dual. I do not know if this is related to
This last adjunct, needless to say, is intended for foreign
the Japanese invasion of Korea at the visitors or residents, just as hiragana assists out-of-towners
end of the fifteenth century.
or anyone else who may know the name of the station they
5- In the sense that Chinese writing itselt want, but do not necessarily know how it is written in kanji,
can be said to be non-phonetic. Actually
Chinese character~ contain
a fact which often causes arduous problems in the usc of the
quasi-phonetic 'operators', hut the telephone directory, for example. The individual Japanese
Chinese language contains no purely
phonetic characters, as any text written
maintains throughout his or her lifetime a continually grow-
in Japanese invariably does (there are ing relationship with the written language, one which
no kanji for such particles dS no, wu. to,
for Western loan words or verb
moves constantly towards a greater mastery of the non-
inflexions). phonetic5 but which can never entirely abandon the phone-
tic, which involves indeed continually turning back to the A SYSTEM OF SIGNS
phonetic, whenever it is necessary to distinguish between
the several possible meanings (and pronunciations) of a
given kanji. Japanese school-children learn successively
four different scripts: hiragana, romiiji, katakana (equiva-
lent to our block printing and derived from the Chinese
characters of kana majiri by a process of subtraction and
geometricalization as opposed to the cursive stylization
which produced hiragana) and kanji. Of course, very few
Japanese ever learn all of the thousands of Chinese charac-
ters which constitute the supposed baggage of a highly
educated Japanese. The minimum number of characters
required of the high-school graduate is I,8so, but every
Japanese, to some extent, goes on learning new characters
all his life (and needless to say, the literary sophistication of
an individual is judged by his or her repertoire of kanji, of a
text by the ratio of kanji to kana or phonetic signs).
This rather long historical resume was necessary to
demonstrate one basic point of great theoretical impor-
tance: the Japanese are the only people in the world who,
for over a thousand years, have practised simultaneously
and in close symbiosis a phonetic and a non-phonetic writ-
ing without taking either as the privileged centre of lan-
guage.6 This is in marked contrast with the experience of
the Vietnamese, for example, upon whom a phonetic
alphabet was bestowed by French priests early in their
colonial history and who soon lost touch completely with
the Chinese script which they too had adopted from their
northern neighbours. (This effect of European colonialism
is of considerable significance, as we shall see.)
What is the relevance to our investigation of this unparal-
leled cohabitation? The answer to this question is facili-
tated by a consideration of one of the basic premises of the
work of the French philosopher Jacques Derrida. 7 6. The earliest Chinese translations of
Derrida's fundamental critique may be said to proceed Buddhist scriptures from sanskrit
involved a phonetic use of Chinese
from this simple observation: throughout the modern his- characters in the transcription of both
tory of the Western world, writing has been regarded as the untranslatable Indian names and
semantically meaningless incantations
passive member of the family of language, a mere transcrip- (see Sansom, An Historical Grammar).
tion of speech, of the logos, regarded as the repository of The fact that this practice never
developed into a full-fledged syllabary
ultimate truth, whilst writing was merely the contingent but died out rapidly points to a
temporal 'form' given to the essential and ultimately divine fundamental distinction between
Chinese and Japanese 'cultures'.
content.
7. I am fully aware of the schematic and
incomplete character of the following
As it has been more or less implicitly defined, the essence of the account, but it is impossible to explore
phone is regarded as immediately next to that which in 'thought' is this subtle body of thought in depth
here. Readers already familiar with
related as logos to 'meaning', producing, receiving, articulating Derrida's writings will, I hope, grant me
and resembling meaning. If Aristotle, for example, holds that 'the their indulgence, while others who find
special interest in these problems will
sounds uttered by the voice ... are the symbols of the moods of refer to the original text (Derrida,
the soul ... and written words the symbols of the words uttered by Jacques, De Ia Grammatologie).

37
GROUNDS, PREMISES the voice' (On Interpretation, I, 16 3) it is because the voice, as
producer of the primary symbols, is in a relationship of essential
and immediate proximity to the soul. As producer of the primary
signifier, the voice is not just one signifier among many. It signifies
the 'mood of the soul' which in turn reflects back, or is a reflection
of, things in terms of resemblance. Between Being and Soul,
between things and affections, there is assumed to be a relation-
ship of natural translation or signification; between the Soul and
the Logos, a conventionally symbolic relation. And the prime
convention, regarded as immediately related to the order of
natural and universal signification, is presumably produced as
spoken language. The written language in this view merely fixes
conventions linking other conventions. 8

This Aristotelian doctrine has been refined and developed

by nine centuries of Christianity and two centuries of
bourgeois rule. It has, however, never ceased to determine
that Western attitude towards language which Derrida
terms logocentrism, demonstrating its decisive complicity
with metaphysics and philosophical idealism - often
through the metaphorical conceit by which the Word
significantly becomes Book, Scripture. This outlook Der-
rida illustrates by a quotation from Rousseau:
'It was as if Nature spread out before our eyes all Her Magnifi-
cence as text for our conversations ... And so I closed all books,
for there is one that is open for all eyes, the Book of Nature. And it
is through that great and sublime Book that I learn to serve and
worship its author' (Emile, Profession de foi). Thus good writing
has always been that which is precisely meant to be understood:
within a Nature or a Natural Law, created or not, but originally
conceived as an eternal presence. Understood and contained,
then, within a totality and wrapped in a tome. The idea of a book is
that of a totalization, finite or infinite, of the signifier; that consti-
tuted totality of signifiers is pre-existent to the book, watches over
its inscription and its signs, ideally it is independent from it. It is
the encyclopaedic protection of theology and logocentrism
against the disruption of writing ... 9
(We shall begin to see the importance of this opposition
closed book/open text in relation to Heian poetry.)

At the same time, phonetic writing- which Derrida takes

care to distinguish, as we shall see, from the 'generalized'
notion of writing (the sense of writing)- has been intimately
linked with the growth of science which empowered West-
8. ibid., pp. 21-2. This theme is also
ern capitalism to exercise an undisputed world hegemony,
dealt with at length in Derrida's Speech for 'the very notion of science was born at a certain stage of
and Phenomena.
writing ... it was conceived and formulated as task, idea,
9· ibid, pp. 3f>- I. project in a language implying a certain type of specific
relationships- structural and axiological- between speech A SYSTEM oF SIGNS
and writing' .10 Another quotation from Rousseau, placed
as epigraph to the first part of Derrida's essay, points to the
essentially ethnocentric character of logocentrism: 'These
three manners of writing correspond fairly accurately to the
three various states under which we may say that men come
together as a Nation. Depicting objects is suited to savages;
signs and words and propositions to barbarians; and the
alphabet to civilised peoples' (Essai sur 1' origine des
langues ). The Japanese, of course, unbeknownst to Rousseau,
were already a 'civilized people' and it is not perhaps
entirely absurd to think that there is some real correlation
between the instrumental requirements of a feudal warrior
class whose objective ambition for some nine hundred
years was national unity organized in a strictly policed
hierarchy around a strong central government, and the
social 'practicality' of a phonetic script. Many war-lords of
the pre-Tokugawa era who contributed to Japan's unifica-
tion could write only kana; it is more than probable that the
time-consuming mastery of the Chinese characters would
have been incompatible with their political and military
activities. And it may even be argued that the secular
practice of phonetic writing contributed substantially to the
preparation that enabled the Japanese to cope with West-
ern science and technology far sooner and better than
peoples whose writing has remained entirely non-phonetic
to the present day.
However, we must avoid the all too familiar temptation
of placing Japan 'on our side'. For the purposes of this
study, the most important single consequence of this
unique double practice is that the Japanese, during most of
their 'civilized' history, had a continual, day-to-day experi-
ence of an absolutely critical linguistic difference which the
peoples of Europe and China could grasp only 'theoreti-
cally' through the study of foreign systems. This difference
is inscribed within their language, where it may not be too
far-fetched to see it as a 'functioning emblem' of that dif-
ference which Derrida, developing an idea of Saussure's,
has taught us to be the intangible principle of language:
If the conceptual part of value is constituted solely by relation-
ships and with regard to the other terms of the language, the same
may be said of its material part; what counts in a word is not the
sound itself but the phonic differences which enable us to distin-
guish this word from all others, for it is they that convey meaning ...
A fragment of language could never be founded on anything but
its non-coincidence with the rest.
Derrida argues from this passage in Saussure's Cours de
linguistique generale (p. 163) that 10. ibid, p. 42.

39
GROUNDS, PREMISES before being or not being 'noted', 'represented', 'figured' ... the
linguistic sign implies originating writing.

Whence the generalization in his work of the notion of

writing in order to demonstrate that logocentrism and a
hypothetical 'grapho-centrism' are equal and opposite
ideologies; both speech and writing derive from 'difference
as the source of all linguistic value.' 11
Whatever the metaphysical perils of this originating
postulate, Derrida's analysis does point up the fact that the
Japanese writing system occupies a privileged middle
ground, nearer perhaps than either the Chinese or the
Indo-European systems to a dialectically constituted level
of reference of languages. It may thereby afford access to
both a linear mode of linguistic representation, such as that
of the West, and to an 'oriental' mode which it is legitimate
to regard, in a theoretical perspective, as a 'practical'
critique of linearity. The linearization of writing and the
linear conception of speech are rooted in the Western sense
of time based on movement in space. Saussure's whole theory
of 'the linearity of the signifier' can, according to Derrida,
be interpreted from this point of viewY That Chinese
writing (and hieroglyphic writing in general) is distinguish-
able from phonetic writing by its (semantic) non-linearity is
a truism. That this difference should correspond to that
between ancient Chinese thought and the main-stream of
the Greco-Christian tradition is not quite so obvious. A
glimpse of this fundamental hiatus is offered by Joseph
Needham:

The implication was that the universe itself is a vast organism, with
now one now another component taking the lead - spontaneous
and uncreated it is, with all the parts co-operating in a mutual
service which is perfect freedom, the larger and the smaller play-
ing their parts according to their degree 'neither afore nor after.'
... The conviction that the universe and each of the wholes
II. ibid., p. 77· composing it have a cyclical nature, undergoing alternations, so
dominated [Chinese J thought that the idea of succession was
12. Derrida adduces in support of this
statement the following quotation from always subordinated to that of interdependence. Thus retrospec-
Saussure: 'The acoustical signifiers have tive explanations were not felt to involve any difficulty. 'Such and
only the line of time at their disposal;
their elements appear one after such a lord, in his lifetime, was not able to obtain the hegemony
another; they form a chain. This because, after his death, human victims were sacrificed to him.'
characteristic is apparent as soon as they
are represented in writing ... The Both facts were simply part of one timeless pattern. 13
signifier, as it is auditory by nature,
unfolds only in time and possesses the Our examination of certain aspects of Japanese literary
characteristics which it borrows from
time (a) it represents an extension and practice will demonstrate that Japan has always resembled
(b) that extension is measurable in only China in this 'indifference' to linear causality: 'Chinese
one dimension: it is a line.' (Cours de
linguistique genera/e, p. 103.) cosmology when it became known to the Japanese undoub-
13. Needham, Joseph, Science and
tedly impressed them by its range and its coherence. It had
Civilization in China. Vol. 2, pp. 2R!H). great influence upon their minds, an influence which still

40
endures. Far Eastern people are apt to hark back to this A SYSTEM OF SIGNS
ancient theory and think in terms of its catalogue for the
guidance of their lives.' 14 In conclusion, I would further
claim that in those areas in which the Japanese mind has
been most productive (not in philosophy or science, but in
the arts and letters), this double experience of linear and
non-linear writing modes may have helped them to radical- 14. Sansom, George, A History ofJapan
ize Chinese thought in their aesthetic practice. to 1334, p. 73·

41
 3. A Boundless Text

The earliest examples of Japanese wntmg convey the

prime social importance which this activity was to
assume, even at its most literary, throughout the forma-
tive years of her history. The Kojiki (712) and the Nihon
Shoki ( 720) were both chronicles designed to legitimize
the reign and power of the imperial house by instituting,
through writing, Japanese mytho-history and in particu-
lar, the divine ancestry of the imperial house. Not only
did writing remain for centuries the privileged instrument
of the dominant political class - a characteristic of the
early stages of all societies - but within that class writing
was to occupy a place in the social order such that it may
practically be said to have been the social equal of speech.
However, it was not until after the seat of imperial rule
had moved from Nara to Heian-Kyo in 794 that textual
practice became in fact a way of life for court and clergy
(i.e. virtually all of literate Japan). It is indicative of the
social concern with language in general that characterizes
this age (and which can certainly be traced to China) that
the very name of Heian-Kyo should be polysemous.
Read as such, the words mean the City of Peace and
Tranquillity, but 'apart from having an auspicious ring
(the latter part of the eighth century had witnessed a
great deal of political strife and bloodshed), the name
combined the first and last syllables of Heijo, Japan's ear-
liest real city, and Ch'ang an, the great T'ang metropolis
on which Heian, like the former capitals, was modelled.' 1
As we shall see, this polysemy will be an essential feature
of Heian literature.
I have already pointed out that Japanese literature
began to flourish about the time of the codification of the
native syllabary, which soon became so dominant in the
novels and poetry of the next few centuries, while the
Chinese script continued to hold sway with comparable
r. Morris, Ivan, The World of the exclusiveness in documents of state, commerce, etc. This
Shining Prince, p. r8. This remarkable
portrait of He ian society was the source
division of labour, so to speak, was related to the radical
of much of the material set forth here. discrimination between the sexes in the upper classes,

42
who considered it unseemly for women to read or write A BOUNDLESS TEXT
kanji. China and her language were still regarded as the
repositories of all true learning (much as were Greece,
Rome and their languages in the West), and of course
too much learning was 'not good' for women. It was
among the idle and often, in fact, highly educated women
of the court that there arose the world's earliest corpus of
novels, as we understand the term today. Foremost
among these was Murasaki Shikibu's monumental Tale of
Genji ( Genji Monogatari). Despite this sexual segrega-
tion, male courtiers also employed the syllabary in the
writing of poems, partly because many were in fact epistles
addressed to women who, in theory at least, could not
read kanji. Another reason was the fact that this poetry
presumably derived from specifically oral tradition
and was, as we shall see, associated with specifically
oral practices. Still another reason was that, considering
the amount of versification required by even the most
trifling social occasion, the syllabary was much more
convenient.
The composition, exchange, and quotation of poems was cen-
tral to the daily life of the He ian aristocracy, and it is doubtful
whether any other society in the world has ever attached such
importance to the poetic versatility of its members .... Upper-
class Heian life was punctuated with poetry from beginning to
end, and no important event was complete without it. Birth was
attended by an avalanche of congratulatory verse; poetry
exchanges were a central part of the formal courting cere-
monies; and, when death approached, the Heian gentleman
would round out his existence with a parting poem 2

And, of course, poetry-writing was essential to the count-

less love-affairs which were among the chief occupations
of that society: upon first glimpsing (or even simply hear-
ing about) a prospective partner, a man would send her a
poem, and she would reply with another. This prelimi-
nary exchange might continue for days or even weeks; if
the gentleman's suit were ultimately successful, a meeting
would be arranged at which the moment of consumma-
tion was preceded by graceful allusions to the themes and
phraseology of the poetry that had led up to it, including
perhaps oral improvisations of new poems. Next came
the obligatory 'morning after poems,' and then for days,
weeks or months messengers were kept busy scurrying
back and forth between the two lovers, often at intervals
of only a few hours, until at length the inevitable poems
of parting were exchanged. And the beauty of the writing
itself was as important as the text: it was quite customary
for a man or a woman to fall in love with a person of 2. ibid., pp. , 9(}--{)2.

43
GROUNDS, PREMISES whom he or she had glimpsed only a sample of his or her
calligraphy. 'Handwriting, it was the general belief,
revealed (or rather expressed) character, breeding, dis-
tinction, and other qualities more clearly than speech,
and it ranked high among the fine arts. ' 3 For it must be
pointed out that this 'reading' of calligraphy had nothing
in common with Western handwriting analysis; it was the
purely aesthetic quality of the writing, its plastic rather
than its personal or 'psychological' qualities which were
being weighed thus, as signs of personal virtues: for good
and evil in this society were equated with beauty and
ugliness.
One significant feature of Heian poetry is that an
overwhelming share of it was written in a single, strictly
codified and remarkably brief form, the tanka: five
periods consisting of five, seven, five, seven and seven
syllables respectively. It is remarkable - and this is but
one instance of that 'boundless' textuality with which we
are concerned - that this alternation of five- and seven-
syllable periods, already present in pre-Heian poetry,
continued to determine the morphology of nearly all
Japanese verse (in both its literary and theatrical forms),
even after tanka had 'given way' 4 to renga, renga to
haikai and haikai to the late Tokugawa lampoons called
senru. And the same will be found to be true of the deci-
sive characteristics of Heian literature as a whole, namely
3· Sansom, George, A History ofJapan polysemy and intertextuality.
to 1334 p. 186. One of the earliest masterpieces of Japanese poetry,
4· Given way only in the sense that it and which pre-dates the installation of the imperial capi-
ceased to be the dominant verse form. tal at Heian-Kyo by nearly a century, already constitutes
Tanka, like renga and haikai, continue
to be practised to this day by millions of one of the most staggeringly sophisticated uses of
Japanese. polysemy to be found in the literature of any language. It
5. The relatively rare (to the tanka) form is 'On Seeing the Body of a Man Lying Among the
of this poem is called choka in this early Stones on the Island of Samine in Sanuku Province' by
period, nagauta when it became, in the
seventeenth century, the foundation of the greatest master of the period, Kakinomoto Hitomaro
the kabuki text. (c. 68o-700). 5

44
Tamamo yoshi 0 the precious land of Sanuki
Sanuki no kuni wa Resting where the seaweed glows like polished gems!
Kunikara ka Perhaps for its precious nature
Miredo mo akanu I never tire in my gazing on it,
Kamakura ka Perhaps for its holy name
Kokoda tOtoki It is the most divine of sights.
Ametsuchi It will flourish and endure
Hitsuki to tomo ni Together with the heavens and earth,
Tariyuka With the shining sun and moon,
Kami no miomi to For through successive ages it has come down
Tsugite kuru That the landface is the face of a god.

Nakano minato yu Having rushed our ship upon the breakers

Fune ukete From the busy port of Naka
Wa ga kogikureba We came rowing steadily till the wind
Toki tsu kaze That rises with the tides
Kumoi ni fuku ni Stormed down from the dwelling of the clouds-
Oki mireba Looking back upon the open sea,
Toinami tachi I saw waves gathering in their mounting surges,
He mireba And looking off beyond the prow
Shiranami sawaku I saw the white waves dashing on the surf.

Isanatori In awe of the terrible sea,

Umi o kashikomi Where whales are hunted down as prey,
Yuku fune no We clutched the steering oar,
Kaji hikiorite Straining the plunging ship upon its course;
Ochikochi no And though here and there
Shima wa okedo We saw the scattered island coasts
Nakuwashi To dash upon for safety,

Samine no shima no We sought haven on rugged Samine

Arisomo ni The isle so beautiful in name.
Torite mireba Erecting a little shelter we looked about,
Nami no to no And then we saw you:
Shigeki hamabe o Pillowed upon your shaking beach,
Shikitae no Using those wave-beaten rocks
Makura ni nashite As if the coast were spread out for your bedding:
Aradoko ni On such a rugged place
Korofusu kimi ga You have laid yourself to rest.

Ie shiraba If I but knew your home

Yukite no tsugen I would go tell them where you sleep;
Tsuma shiraba If your wife but knew this home,
Ki mo towamashi o She would come here searching for you,
Tamahoko no But knowing nothing of the way-
Michi dani shirazu The way straight as a warrior's spear -
Oboboshiku How must she be waiting,
Machi ka kou ran How anxiously longing for you,
Hashiki tsumara wa. She the dear one you called wife.

45
GROUNDS, PREMISES ENVOYS
Tsuma no araba If your wife were here,
Tsumite tagemashi She would be out gathering your food,
Sami no yama She would pick the greens
No no e no uhagi From the hillslopes of Samine-
Suginikerazu ya. But is their season not now past?
Oki tsunami So you rest your head.
Kiyoru ariso o Pillowed on the rocky spread-out bedding
Shikitae no Of this rugged shore,
Makura to makite While the furious, wind-driven surf
Naseru kimi ka mo. Pounds ever in from off the sea. 6

In order to give some small idea of the essentially polysem-

ous nature of this text, I shall quote two fragments of the
extensive analysis given by Brower and Miner.

The most remarkable thing about this syntax is the way in which it
plays upon certain inflectional endings of verbs and adjectives in
an ambiguous fashion. In eleven instances there are constructions
that may be taken as finite and so conclusive, but at the same time
as attributives to what follows ... In short. what Hitomaro has
really done is to present us with a poem that consists of a single
sentence in which every successive syntactic element is governed
by what precedes, all the way back to the beginning of the poem.
More than this, the last word and the particle tsumara wa (topic:
'as for [your] wife', with the suggestion in the endearment plural
ra offJmily, homeland, and all associated with her), arc not only
governed by everything that precedes but also are the beginning
of the 'main' sentence of the poem. a sentence that stops with no
more than this topic. Such syntax is unimaginable in English, but
something of its effect can be gained by beginning with the last
word of the Japanese and working back in translation, without a
break, to the initial line of the poem .... Primarily what interests
[Hitornaro] is not so much the actors as the actions, not so much
the issue of responsibility as the integrated nature of the process of
a complex experience. Consequently, there arc verbs in abun-
dance and noun images to create the scene vividly. And the syntax
is so organized that in spite of the numerous semi-stops, the stops
arc transformed into attributives for the noun-clauses that follow.
6. Brower, Robert H. and Miner. Earl,
The skill involved is prodigious. as one can sec from the transition
Japanese Court Poetry, pp. 97-H. If I between the first and second section. The first section ends with
have quoted this beautiful poem in full,
the verbs tsugite kuru forming an idiom something like 'transmit-
it is only partly as an illustration of the
sophisticated construction~ of ancient ted'. The effect is to say that it has been transmitted that the land
Japanese poetry. For here also is one of
of Sanuki is divine. But kuru is also used separately, apart from
the earliest recorded manifestations of
an approach to narrative which is a the idiom and attributively, for the next clause. Here kuru
constant of Japanese literature and
which was to reach its final and by no
achieves more of its usual sense as an active verb in a clause
means least significant avawr in the something like 'coming to the harbor of Naka'. as if one could take
cinema oft he 19)0S, which also 'revived'
snmc of the structural ~tratcgies of that
in English the '-mit' from 'transmit' and usc it in its root Latin
early poetry. sense of 'send' as part of the relative clause. There arc, as we have
said, eleven such double-functioning verbs and adjectives in the A BOUNDLESS TEXT
poem, and the result is that by such hypo taxis, Hitomaro manages
to integrate all the elements of the poem into one continuous
poetic and linguistic process, into one experience. 7

This procedure became standard practice throughout

He ian poetry under the generic name of kakekotoba, or
pivot-word, 'a rhetorical scheme of word-play in which a
series of sounds is so employed as to mean two or more
things at once by different parsings. For example, nagame
used to mean reverie (nagame) and long rains (naga ame).' 8
Like the alternation of five- and seven-syllable periods, it
has remained a constant component of Japanese verse-
forms. This feature of Japanese poetry shows the extent to
which the 'Japanese mind' rejects linearity and the 'trans-
parency of the signifier' which has dominated both Western
thought and art since the eighteenth century. The language
process is constantly at work on the very surface of these
poems.
In the West it was in the eighteenth century, which
witnessed the rise of the bourgeoisie and the reassertion of
logocentrism and the emergence of an ideology of rep-
resentation suited to the needs of the bourgeoisie, that the
masking of the process of the production of meaning
became as important, on its own level, as that of the process
of production of goods. This ideology continues to domi-
nate our notions of representation to this very day. As we
shall see in the next chapter, it is this ideology of the
transparency of the sign which dominated the emergence of
the Western film from its 'primitive' stage. Conversely, no
Japanese artistic practice, from the earliest known poetry
under discussion up through the theatre and literature of
the Edo (i.e. the Tokugawa) period, ever subscribed to
such a notion. This observation may be regarded as central
to my thesis. The poem by Hitomaro reproduced above is
'about language', it is about the process by which we make
meaning and in the original Japanese can scarcely be called
representational at all, so compressed is the syntax.lt is this
inscription of the signifying process in the 'text' which is
such an essential characteristic of the traditional Japanese
arts and which was to influence the development of Japan-
ese cinema in this century.
Another important feature of Heian poetry for the pur-
poses of this study is that to which we have already referred
as intertextuality, a practice which, in various ways, totally
contests the myth of the closed text and the concomitant
notion of originality.
This concept of intertextuality has broad implications 7· ibid., PP· r43-6.
for, as we shall see, it ultimately includes the 'text' of the s. ibid., p. 5 o7 .

47
GROUNDS, PREMISES society in which it was operative. The practice of relating
one poem to another by allusive variation (honkadori)~
quite naturally led to the composition of poetic sequences,
first in connection with the contests (utaawase) in which
poems were composed upon common themes in order to
establish grounds for comparative judgement, and in the
imperial anthologies, of which twenty-one wac compost:d
between 905 and 1433. These anthok>gies are unique in
world literature, for they are not merdy colk.:tions of what
the compiler considered to be the 'best verse of the reign'.
The poems are arranged (and in many instances, one sup-
poses, chosen) to fit into successive and/or overlapping
patterns of various sorts. The supreme sophistication
reached hy this technique is generally regankd to be found
in the Shinkokinsh{t, commissioned by the ex-emperor
Go-Tuba in 1201.
Ados<' n:ading ,,f the Shinkokinslul WllUld giv,· us .tlmosi innum-
.:rable examples to illustrate the tcdmiqu..:s. A seq Hence nf travel
pt>ems t,x:S96-·907) shows that the arrang..:ment for progression
~tlone is based in part Pn what might be called a plut ~~i•tce there is
a n>n<·atcnation ,,r situati,ms), \\ lnd1 take~ us lHl aVO\ .tge fi'lllll a
ft>n:st. ttl tlw .:oasr, Ill the S<'a, ttl ( 'hina, back ll• Ltpat· and soon
and 1111. Blll, tlll'rnlibk as it Sl'<'lllS, this hnok ts at lh<' same time
also ordered as a kind ,,r history of Japam'S<' poetry. lt .1pens with
poems from the early literary period and moves down t,J th..: poets
of Go-Toba's own time. Or, to take another example, tlte first four
poems on love in Book xt (990-993) show the tec!miques of
ass,Kiatiunal development: the unity of the age of the poets wlw
wrote the poems, in the distant past; a succession from mountains
to hills to uplands tu plains; an even subtler sequenn of images
from snow to waterfalls to grain; and all of these con tamed within
a progression of geographical distance and further umfied by the
point of view of the speaker of the poems. 111

The organization of these anthologies, which is both inter-

textual and polysemous, manifests the fundam(·ntal need
for refined patterning in all human activity- from religious
sculpture and architecture to the arrangement •>f t1owers
9. "... an echoing of an older poem or and food- which has informed Japanese culture for cen-
poems, not just to borrow material or
phrasing, but to raise the atmosphere-
turies. This impulse may be said to have received its defini-
something of the situation, the tone, the tive impetus in the mid-Heian period under the Fujiwara
meaning- of the original.' Brower and
Miner, op. cit., p 14
regents.
The anthologies were marked by another feature which
10. ibid., p. 324.
is also of interest to us here: with increasing frequency,
1 1. They were conse4ucntly separated the poems were prefaced" by explanatory headnotes
from each other, which appears to be
one of the reasons why the very
(kotvbagaki), 12 which generally amounted to a few lines
existence of these sequences was not about the situation in which the poem was comp,Jsed. lt is
re-discovered until this century.
said that this practice was often dut: to certain 'obscurities'
12. Etymologically: 'framed words'. in the text, deriving from the private nature uf ct:rtain
allusions. From our viewpoint, however, it primarily A BOUNDLESS TEXT
embodied the fact that poetry was an activity within a
context, either private or public. This awareness is in sharp
contrast with traditional Western attitudes which tend to
close off the text, not only from other texts but, above all,
from the social 'text' in which it is nonetheless inevitably
inscribed. This presence of the context is a permanent fea-
ture of the Japanese difference, in both He ian literature and
modern film practice.
After the first anthologies, a new literary form, clearly
derived from this poem-headnote relationship, began
to emerge: these were collections of 'poem-tales' (uta-
monogatari) each illustrating or recounting the production
of one or several poems. The most famous example of this
transitional but important form was the Tales of lse, (Ise
Monogatari) ascribed to the poet Ariwara Harihira
(825-80) and purportedly describing his life and loves. The
following episode is representative.

Once a man and a woman lived in a remote district. The man,

saying that he had to go in order to serve in the Palace and
regretting to part from his wife, went away. Thereafter, three
years elapsed and he did not come back, so that his wife had grown
tired of waiting, but, just when she had promised to another man
who had made advances to her in a very kind way: 'Let us meet
to-night', that man came back. He knocked on the door, calling:
'Please open this door,' but without opening it, she just composed
a poem and had it brought to him:
Tired of waiting
Three of these years
Which always renew themselves
I am to try a new pillow
This very night.
As she had just sent word to him, the man returned the following
poem:
Just like I have been to you
Through the years
As different as different kinds of bows
You should be affectionate to him
When he was about to go after these words, the woman recited:
No matter whether others draw my sleeves or not
- Minding of the drawing of a bow of Catalpa wood
Always
Has my heart
Been drawn to you ...
But the man had gone back to the capital. Although the woman,
feeling very sad, started after him and pursued him, she could not
overtake him, and at last she had fallen down at a place where
there was a spring. On a rock which was there she wrote with the

49
GROUNDS, PREMISES blood of a finger the following poem:
Unable to hold back
The man who was departing
Without answering my love,
Now my body
Seems to have faded away to nothing.
Having written these lines, she passed away. 13

Here the notion of narrative is consubstantial with a pre-

sentation of the process of literary production. The text as a
whole is also a reductive dramatized model of the life-
pattern of the He ian court in so far as it hinged upon textual
practice and sexual relations.
The next stage in this amazingly integrated development
was the fictional tale (tsukiri monogatari), of which the
six-volume Genji Monogatari is the most famous example,
although the term does also designate very brief texts, often
little longer than that quoted from the Ise Monogatari.
Again, in these tales and in the diaries which no courtier of
the day failed to keep, poetry is constantly present in its
social function; present, as well, are all the modes of inter-
textuality and of polysemy with which we are already famil-
mr.
One example, which combines the two, is taken from an
anonymous text, Ochikubo Monogatari, a title which is
sometimes rendered inadequately into English as The Tale
ofthe Room Below. A court official, the Shosho, is secretly
courting the Lady Ochikubo, who is subject to constant
persecution by the Kita no Kata, the wife of another court
official, in whose house the Lady lives in Cinderella-like
humiliation and exploitation.

Hardly knowing what she was doing, the Lady began to pleat the
folds of the hakama but the Shosho caught hold of her skirt and
urged her to go back to bed. Smiling, she went back to bed, for
there was nothing else to do.

'How hateful she is! [referring to the Kita no Kata]. Do not do any
more sewing. Let her become more angry still. What horrible
language she uses towards you! Has she been as abusive as this all
these years? How have you been able to endure it?' he exclaimed.
'I am a flower of the wild-pear-tree,' the Lady quoted in answer. 14

The translators, who are constantly obliged to explain allu-

sions, indicate, somewhat enigmatically, that this is a quo-
13. Vos, Frits, A Study of the tation from the following poem:
Ise-Monogatari, p. 191.
Although .in this house
14. Whitehouse, Wilfrid, and
Yanagisawa Eizo (trans), Ochikubo
I am able to find nought
Monogatari, p. 6r. But care and sorrow;

50
 Yet it seems that nowhere else A BOUNDLESS TEXT
Can I find a hiding place.

Though the aptness of the poem is clear, the translation

contains no mention of 'a wild pear tree.' As the translators
explain, 'the reference to the "pear-flower" disappears
when the meaning of the poem is considered.' 15 Or rather,
one of its meanings, since clearly we are dealing with an
oblique instance of polysemy.
In connection with Ochikubo Monogatari, it is interest-
ing to point out that in the novels of that period, almost
none of the characters are ever designated by their names
but rather by their titles or other terms which situate them
in their social functions. 'Kita no Kata' and 'Shosho' are
official court titles, while 'Ochikubo no kimi' (translated as
Lady Ochikubo) means the 'Lady of the Lower Room', a
mock-title with a slightly pejorative nuance to 'lady'. This is
a highly significant practice, to be compared with the long-
standing Japanese custom (maintained even today among
professional practitioners of the traditional arts) of chang-
ing a person's name according to his or her social situation.
The ShOguns of the Kamakura period (II85-1333), for
example, were often called by as many as four names
during their lifetimes: as a child, then perhaps as Shogun-
to-be, as Shogun proper and then, in the event that they
retired before death or were removed from office by the
regent, as monk, priest or whatever. This is a significant
instance of the Japanese attitude towards the notion of the
person or subject: in the West, the name is co-extensive
with the self, it is the mark of the soul. For a man to change
his name is a serious sacrifice of identity, to be consented to
only for very serious reasons (to escape police detection or
racist persecution, for example) while for a woman that
'sacrifice' has always been the symbol of her subordination
to man, in other words her voluntary abandonment of a
vital part of her identity. In Japanese history, there was no
such attitude; a name, it was felt, was more properly
attached to a function than to a physical person, and when
the functio~> changed, it was normal that the name should
change too.
It must not be imagined that the advent of a more 'realis-
tic' form of literature, the tsukuri monogatari, meant a
slackening of the 'structural vigilance' which characterized
the other forms of the period. Ivan Morris points out sev-
eral recurring formal devices worthy of Flaubert which give
the huge text of Genji a structural coherence which did not
appear in the Western novel until the late nineteenth cen-
tury. Repetitions of situational patterns, 'sustained imag-
ery', (the repetition of a single central image in both the 15. My italics.

5I
GROUNDS, PREMISES narrative passages and the frequent poems), are two of the
most representative, and most clearly related to the work of
the signifier in poems and anthologies. 16
Any account of the intertextuality typical of He ian litera-
ture must include its ties to nearly all Japanese literature for
the seven centuries that followed the political and cultural
decline of the imperial capital. As the tanka began to lose
its vitality in the thirteenth and fourteenth centuries, there
arose a new form, linked verse (renga ). This was an obvious
derivation from the poetry contests and anthologies: sev-
eral poets would gather for the purpose of composing a
collective poem according to very strict rules. 17 Here we are
dealing with perhaps the most radical effacing of the
'creator' to be found in any post-primitive practice, for
indeed the avowed aim of these poets, as expounded by
Yoshimoto Nijo ( 132o-88) and as paraphrased by the con-
temporary Japanese scholar Ueda Makoto, was the effac-
ing of the creator-subject and the identification of the pro-
cesses of reading and writing.
The poet and the reader are nowhere more closely related than in
linked verse, because here the reader of one stanza may become
the poet of the stanza immediately following. To compose a stanza
for linked verse the poet must first try to become a perfect reader
of all the stanzas preceding his; he has to put himself in the
positions of all his fellow poets sitting around him. One obvious
consequence of this is a demand for the poet to suppress his
individuality: the poet must work within the framework set by
other poets as well as by the contemporary rules of poetic compos-
ition. In this respect, the theory of linked verse stands directly
opposite from that of lyric poetry; instead of speaking out his
personal emotion, the poet must dissolve it for the sake of the
team of which he is a member. 18
This demonstration of the profound equivalence of reading
and writing speaks directly to modern artistic practice and
theory. Eisenstein's materialist conception offilm-making,
as described in the following statement, can be seen as
related to this attitude: 'It draws the spectator into a crea-
tive act in which his own personality is not dominated by
16. See Morris, op. cit., pp. 277ff.
that of the author, but fully develops in harmony with the
author's conception, just as the personality of a great actor
17. Some of these bear remarkable fuses with that of a great playwright in the process of
similarities to the privative rules
governing orthodox serial music, in creating a classical image on the stage.o~ 9 We will further
particular the prohibition to use certain consider this attitude at a later time.
words more than once or certain others
more than twice. In connection with renga it is also worth noting here that
besides its multiple affiliations with Heian literary tech-
18. Ueda Makoto, Literary and Art
Theories in Japan, p. 38. niques, it is intimately related to a three-fold module which
I g. Eisenstein, Sergei, Notes of a
gives the no theatre its characteristic form. Traditionally a
Film-Director, p. 78. no performance lasts a full day and involves a complete set
52
of plays. Each play, each section of the play and each A BOUNDLESS TEXT
sequence of plays is divided into three sections called jo
(introduction, 'quiet in mood and tempo'), ha (central sec-
tion, 'with a lighter mood and leisurely changes of pace')
and kyu (finale, 'with heightened rhythm and forceful
impact'). 20 This module, like so many others to be found in
the 'text' of Japan, is remarkably ubiquitous, for not only
does it condition the progression of renga, of certain forms
of classical music and of the no, but it was also observed in a
non-competitive form of football! And of course the no
play itself, despite its 'humble' origins in the early medieval
popular theatre known as sarugaku, 21 is also directly
related to the textual practices of He ian court poetry (e.g.
the five/seven pattern, the allusive variation, the pivot
word) and to a fund of traditional lore on which all Japan-
ese narrative arts (including the cinema) draw so exten-
sively.
The 'novels' of Ihara Saikaku (1642-93) (and particu-
larly The Japanese Family Storehouse) provide a further
extension of the He ian module. They consist of apparently
aimless, 'uncentred' short-stories, whose sections are
ordered according to techniques derived from a later
development of linked verse (which also produced the
haikai, of which Saikaku was a master). Saikaku's work will
enter into the examination of narrative structures em-
ployed in certain films of the 1930s.
With this admittedly skeletal account, I hope to have set
forth four basic contentions which are closely related and
which will recur in one form or another throughout this
study. Firstly, tradition inclines the Japanese to read any
given text (and this may also be a film, as we shall see) in
relation to a body of texts. Secondly, the sacrosanct value
placed on originality, the taboo placed on 'borrowing', on
'copying' in the West are as utterly foreign to Japan as are
Western 'individualism' and the primacy of the person or
subject. Thirdly, the linear approach to representation is
not a privileged one. Finally, the precedence given 'con-
tent' over 'form', or rather the hypostasis of meaning to the
detriment of its production, is a specifically Western
attitude. It has informed all Western methods of analysis,
explanation, reading or interpretation; it has also been
imported into Japan to fill an undeniable theoretical and
instrumental void. It has, however, no place in any artistic
or other practice that can be identified as specifically
Japanese.
This preliminary exposition is confined to Heian litera- 20. Ueda Makoto. op. cit., p. 45·
ture and its descent. Other, more recent artistic practices
21. Literally 'monkey music', possibly
have made distinct and specific contributions which have referring to the acrobatic and comic
informed, directly or indirectly, film-making in Japan. mimicry of sarugaku actors.

53
GROUNDS, PREMISES Heian writing is linked with these other practices (scroll
painting and 'picture-poems' are especialiy striking
instances). Many of these- especially in painting, theatre,
architecture and garden design - will be referred to in
connection with the films of specific periods or directors.
The 'boundless text' generated by Heian society contains,
explicitly or implicitly, all the basic theoretical challenges
that Japan offers Western thought and practice. As such, it
constitutes a proper introduction to the problems implicit
in any coherent study of Japanese cinema. Although many
ofthe modern theoretical concepts touched upon here have
not yet been sufficiently delineated, especially for readers
unfamiliar with their Western European context, it is my
hope that through repeated illustration in the pages to
follow, remaining obscurities may be progressively
clarified.

54
Part 2 A Frozen Stream?
4. A Machine Appears

Motion picture production began in Japan in 1898, just

thirty-one years after the Meiji Restoration. The urban
population had been swept by a mass emotional response
to the ruling oligarchy's proclaimed recognition that the
country's future independence was contingent upon her
achieving the status of an industrial power on a footing
with the great nations of the West, and to their con-
comitant policies of accelerated assimilation of Western
techniques. This had taken the form of a craze for Wes-
ternalia which led even reflective men to advocate such
radical abandonments of national identity as the
wholesale adoption of the English language in lieu of the
native tongue and script, or the promotion, on a nation-
wide scale, of mixed marriages between Japanese men
and Western women to 'fortify the race'. The peak of this
craze had already begun to pass, especially after Japan's
easy victory over China in the war of 1894-5, which
greatly strengthened the position of the advocates of the
'preservation of the national heritage'. The very real
enthusiasm which nonetheless greeted this latest novelty
from overseas may still be considered as part of that first
great elan towards things Western which did not com-
pletely subside until the beginning of the second war with
China early in the 1930s. At the same time, as I hope to
show, the cinematograph struck a number of fundamen-
tal chords in what is termed the Japanese sensibility.
Japanese film-production developed so swiftly that by
1909 four companies were in operation and, between
1909 and I9II, Makino Shozo, 'the first man to deserve
the name of director in the Western sense of the word','
together with his phenomenally popular star, Onoue
Matsunosuke, turned out no less than r68 films (mostly
one-reelers); their annual average for the next decade
was nearly one hundred, with the length of the films 1. Richie, Donald, and Anderson,
gradually increasing as well. With respect to quantity and Joseph L., The Japanese Film, p. 31.
length, then, Japan kept pace with the West; when one Unless otherwise stated, all factual
information on early Japanese cinema is
remembers that Louis Feuillade, a very prolific French drawn from that source.

57
A FROZEN STREAM? director of the same period, made 'only' some 700 films
between 1906 and 1923, it becomes fairly probable that
eventual comparisons between early Western and Japan-
ese cinema are grounded in comparable bodies of film
production. We are not dealing with an industry of mar-
ginal status such as that of all other non- Western cinemas
during what I shall refer to as the Primitive (1894-1908)
and Formative (1909-19) periods of Western film history.
Japanese film-making strove, and in some respects suc-
cessfully, to 'stay abreast' of the West. This was a goal
which the country had set herself in all fields of industry,
but here, for the first time, it was immediately feasible,
since her entry into the field coincided almost exactly
with the birth of cinema in the West.
In 1909, the Nippon Katsudo Shashin (Japan Cine-
matograph Company) was formed by the merger of the
four production companies which had developed since
the turn of the century. In this they were taking their cue
from the Pathe-Biograph-Edison ... cartel in the West,
but also from the general pattern of concentration estab-
lished from the very outset by the architects of Japanese
capitalism. By 1912, this Japanese trust was operating
four studios and seventy permanent theatres throughout
the country. The majority of film-showings, however,
were still being organized in non-permanent settings, as
they were in the West.
A programme of films by Louis Lumiere, including the
pioneer Entree d'un train en gare de La Ciotat (1895),
constituted the first public film-showing in Japan in 1897
(though the Edison Kinetoscope had been imported
three years before). More important for our purposes,
perhaps, is the fact that these screenings, and those of
Edison's Vitascope a few days later, constituted the first
presentation in Japan of a Western 'performing art' in its
original form, just as the West had produced it and as
Western audiences, at that very moment, were viewing it.
The Meiji period had witnessed an influx (or rather a
burgeoning) of 'Western styles' in theatre, the novel and
painting. The visual arts of the West were becoming
directly available, in reproduction, to those Japanese who
were in a position either to appreciate them as connois-
seurs or to draw upon them as practising artists. The ear-
liest 'Westernization' of the novel in Japan was based on
a far more limited familiarity with Western practices.
Understandably, only a few texts were selected for trans-
lation according to criteria determined by the peculiar
needs of Japanese society in a period of vertiginous
change. And these texts were translated into a language
so radically different from those of the West that the
Japanese versions are hardly translations at all, but A MACHINE APPEARS
rather very free adaptations. 2 The late 188os, however,
brought a completely new approach to written prose,
more suited in particular to Western literature in that it
involved the introduction into the written language of the
constructions and vocabulary of spoken Japanese. This
was certainly one of the most far-reaching effects of this
period of Western influence, and it is significant of the
impact of Western modes of representation on those of
Japan.
A new form of theatre, known as shimpa, arose after
r888; it soon came to rival, and briefly to eclipse, the
popularity of kabuki. Shimpa purported to be a 'West-
em' form, which no doubt explains its having played a
direct role in the constitution of the repertory and the
modes of representation in early Japanese cinema. How-
ever, it is important to understand that it was developed
mostly by non-professionals who had practically no
first-hand knowledge of the Western theatre. Shimpa was
a fantasy creation, comparable in many ways to Kafka's
portrait of Amerika; understandably, it bore little if any
resemblance to its Western model.

These men conceived the Western drama to be couched in col-

loquial language, as opposed to the stylized language of the
kabuki plays, to be concerned largely with the contemporary
scene, and to be acted not in the strictly designed movement of
the kabuki but with an imitation of the movement of life. Dur-
ing the nineties, shimpa enjoyed such a wide vogue that even
the foremost kabuki actors of the day felt impelled to play
shimpa roles, and Morita Kanya, of an old kabuki family, estab-
lished a shimpa theatre. With this, the kabuki began to influ-
ence the amateur shimpa performers, and their acting tech- 2. See Sansom, George, The Western
niques came more and more to resemble those of kabuki. Soon World and Japan, pp. 403-4; and Ernst,
Earle, op. cit., p. 25r. As for the works
shimpa showed little difference, either in mode of production or selected for translation, these were
in subject matter, from the plays of the traditional kabuki per- generally either chosen for their
compatibility with the traditional forms
formance. The roles of women in these 'realistic' shimpa pieces of Japanese literature (The Bride of
were played on the same stage during the same play by both Lammermoor, Les A ventures de
Te/emaque) or for what was conceived
men and women. 3 as their usefulness in understanding the
problems of the day, as raised by the
One of the typical and most picturesque ongmators of accelerated process of 'Westernization'
(the non-fiction best-seller of the period
shimpa was Kawakami Otojiro, former policeman turned was Self Help by one Samuel Smiles, a
'revolutionary', who came to the theatre through comic primer for coping with 'modernity')
while the literary successes included the
story-telling (rakugo ). He began in 1888 to produce and novels of Jules Verne, Michel Robida's
act in the new genre but it was not until 1899 that he fancifully illustrated vision of Le
Vingtieme siec/e, and for reasons no
went abroad to glean some first-hand knowledge of the doubt related to the success of both
Western theatre. SelfHe/p and Les A ventures de
Te/emaque, two successive translations
of Robinson Crusoe.
Upon his return, he staged a production of Hamlet in which he
played the leading role. The legend exists that in one scene of 3· Ernst, Earle, op. cit., p. 249.

59
A FROZEN STREAM? the play, Kawakami, as Hamlet, made his entrance to the stage
riding a bicycle. Legend or fact, the story embodies a good deal
of truth about Japanese conceptions of Western drama at that
time .... Hamlet on a bicycle was not incongruous to the audi-
ence of that period, for both Hamlet and the bicycle were new
and foreign and therefore logically belonged together. ... In
short, shimpa was based primarily upon what men of such
dubious knowledge as Kawakami conceived Western drama to
be, expressed throughout in terms of Japanese life and
attitudes. 4

It is a very particular type of transformation which pro-

duces Hamlet on a bicycle, a process typical of Japan's
habitual reading of Western artefacts and techniques: one
may question the judgement that the 'dubious knowledge'
of men such as Kawakami was responsible for this failure to
reproduce the Western modes of theatrical representation
(a failure which, as Ernst points out and as today's visitor to
Japan may easily observe, remains, in many respects, a
reality). This matter might be clarified by comparison with
the evolution of the cinema in Japan since, as we have seen,
for the first time the Western model was present in its
authentic, original form for all to see, spectators ('readers')
and film-makers ('translators') alike.
Before analysing the gradual but ineluctable and far-
reaching divergence between the dominant practices in
Western and Japanese film industries, we must open
another parenthesis, a rather lengthy one, in order to
establish the salient features of the 'evolution of film lan-
guage' in the West during the period 1894-1919. Funda-
mental differences between the cinemas of Japan and the
West are, by the latter date, clearly visible, albeit in
embryonic form. However, unless we understand, if only in
broad outline, the process by which nineteenth-century
ideologies of representation came to determine the rep-
resentational modes of Western film, we shall not under-
4· ibid, pp. 25Q-I. stand the origins and development of these differences.

6o
5. A Parenthesis on Film
History

The earliest motion picture ventures of Edison and his

associates in the United States and of the Lumiere
brothers in France objectively introduced the antithetical
directions in which Western and Japanese films were to
move, predominantly, during the next fifty years. 1 They
also prefigured the basic conflicts which were ultimately
to develop in the cinema of both Japan and the West,
considered separately.
The history of cinema has generally been interpreted
as a development of the alternatives proposed by the
pioneer work of Lumiere and MeWes. These two alterna-
tives are thought to form the basis for the distinction
between documentary and fiction film. I regard the work
of Melies and Lumiere, however, as two aspects of
the same phenomenon. Conversely, the contradiction
between the films shot by the Lumieres and their
cameramen, and some of those produced for the Edison
company during the first few years by Dickson and Raff
and Gammon is I believe absolutely fundamental.
The first relatively successful attempt to develop the
technology of talking motion pictures came out of the
Edison laboratory. It took the form of the kinetophono-
graph of r8g5. It was 'total reproduction of life' that
I. Friese-Greene in England and a host
interested Edison. 2 It was no accident that the first of others are not discussed here.
close-ups came out of that black-walled structure in Edi- Considering, however, the importance
which the films of France and the
son's garden. And his talking pictures were in advance by United States were to assume during the
some twenty-five years. Primitive and Formative periods, it
seems reasonable to pay special
Edison sought the reproduction of perceptual reality as attention to the implications of the
a whole, of 'life'. The silent reproduction of perceptual implicit attitudes of these two pioneer
figures.
movement seems to have interested him rather little. The
orientation of his research perfectly reflects an essential 2. I am fully aware that Edison's
personal responsibility for the film work
aspiration of the bourgeoisie with regard to representa- that took place under his name has been
tion. Auguste and Louis Lumiere, on the other hand, shown to be slight indeed. What is at
stake here, however, is not a personality
were still the direct heirs of the pioneers Muybridge and at all but the characteristic attitude of a
Marey, whose work was motivated by an essentially sci- class, incidentally- and contradictorily
- reflected in this or that individual,
entific aspiration to analyse movement. This in part manifest attitude, but above all
explains why, in their earliest period, they approached inscribed in the films themselves.

61
A FROZEN STREAM? their invention as though it were a scientific toy - which
was to some extent how the cinema was viewed on the
fairgrounds of Europe for over a decade to come. Edi-
son, on the other hand, from the very start, clearly saw
his Kinetophonograph as a fulfilment of 'the age-old
dream of man', the perfectly illusionistic reproduction of
the human likeness and voice, i.e. the bodily envelope
and its souU This apparatus, with the viewer's isolation
(via headphones and eye-piece) and his submersion
thereby in diegetic space-time, is an astounding pre-
figuration of the achieved illusionist rapture of the 1930s.
Edison's personal conviction that the ultimate cinematic
goal was the filming of opera, is a significant extension of
the Gesamtkunstwerk ideology.
The neologisms coined by Edison (Vitascope) and the
Lumiere brothers (Cinematographe) are also emblematic
of their antithetical positions: a 'vision of life' as opposed to
'an inscription of movement'. Nothing better illustrates,
however, the objective ideological difference between
their earliest ventures than the actual conditions in which
they were carried out: Edison and Dickson barricading
themselves in their black-walled, sound-proofed Black
Maria, which prefigures the sound-stage of the 1930s and
after; the Lumieres setting up their camera in front of their
own factory gates a few minutes before the end of the
working day. It is not simply because they were filming
'things as they are' that the Lumieres' method implied a
world-view crucially opposed to that inscribed in the films
of Dickson, Raff and Gammon et al. After all, the bar and
barber-shop scenes reconstituted against the black back-
ground of Edison's tank are in their way strongly 'realistic'.
3· The development of the work of
Demen}', Marey's assistant, who began
The 'greenhouse' constructed by Melies in Montreuil, on
by prolonging his master's analysis of the other hand, was also designed to exclude the world of
movement, is emblematic of the
contradictory forces at work in this
contingency (as was Edison's 'studio'), but it was in order to
period, for he later became the staunch construct a world as radically and avowedly artificial as
promoter of the talking portrait, possible; his aim was not to exclude elements which might
destined in his mind to supplant the
photographic portrait in both family and impinge on the 'vision of life', but simply to prevent any
public archive. The drive for a greater
'totalization' of the signs of reality was
intrusion on a perfectly ordered world of artifice. In short,
very strong during the first ten years, as the opposition studio/location is not in itself fundamental.
witness Gaumont's steadfast attempts
to impose synch sound and the English
It is true that the Lumiere brothers, in their earliest
experiments in colour cinematography. period, regarded the motion-picture camera/projector/
The various efforts to create a 'total',
enclosing cinema, to produce the
printer which they had designed as particularly well-suited
moving image as environment were to recording images of the real (and especially the distant)
especially significant of those world. For some months they treated their invention as a
contradictions: totally non-linear in
effect (spectators could see no more scientific development, a valuable instrument for research.
than one portion at a time of the
panorama surrounding them, and were
It clearly seemed most natural for them at that time to set
free to look wherever they pleased}, up their apparatus in some well-chosen place and then
these were nonetheless naive attempts
to achieve what Griffith and others
'leave it on its own', so to speak. In that sense, each of those
ultimately achieved through editing. early 'documentary shootings' - and there were several
hundred of the sort- may be likened to a scientific experi- A PARENTHESIS ON
ment, in which the organisms or substances under study are FILM HISTORY
allowed to interact with complete freedom within the
specific confines of set conditions. Of course, the Lumieres
were not engaged in any actual programme of research,
except insofar as they developed details of their technology
and filming techniques as they went along. They, and the
cameramen who followed their 'reporting' techniques,
were first demonstrating their 'toy', and later exploiting an
attraction (far more profitably than Edison did for a while).
The Lumieres as individuals were not, in any sense, above
or outside the ideology of their class (Georges Sadoul has
shown how strong was the class content of their first films). 4
Yet one may postulate that their socially and historically
determined 'scientific reflexes' neutralized for a time cer-
tain aspects of that ideology, so that they inaugurated
'spontaneously' the representational model which domi-
nated the budding industry for the next ten or fifteen years.
At this point, it is enlightening to observe the extent to
which the work of the Lumieres and others who best
characterize the Primitive mode of representation, resem-
bles some of the most advanced of contemporary film-
making, that which is most radically and explicitly in con-
flict with the cinema typified by Hollywood and prefigured
in part in the Black Maria.
For example, Warhol's famous quip 'I want to be a
machine' objectively echoes the Lumieres' attitude
towards the process of recording movement on film, just as
their earliest practice is, in miniature, that of Warhol's
pioneering period.
When Peter Kubelka, Kurt Kren and others arrange to
show two or more times the self-same images, their practice
recalls that of the Lumieres' (and also Edison's) projection-
ists screening several times over those first reels of film to
the same unflaggingly attentive audiences, whose 'naive
eye' told them that they did not 'see everything the first
time', that the second time around the film would still be
new, who did not have a 'linear response' to the film
experience, did not as yet demand (or had not yet been
trained to expect) a chain of signifiers on the Saussurian
model, in which each new link would necessarily devour the
previous one.
Finally, the Lumieres' spatially non-centred street scenes
elicit a free-floating scan as opposed to the increasingly
ordered and guided gaze that Griffith and others were to
introduce. Their work was temporally non-centred as well,
devoid as it so often was of beginning or end. Here, the
modernist ramifications are evident. 4· Histoire generale du Cinema, Vol. r,
While D. W. Griffith certainly played a crucial role in the p. 214.
A FROZEN STREAM? systematization of the rules of 'illusionist' editing to which
his name remains understandably attached, he was pre-
ceded by others, in particular the English pioneers (Smith,
Williamson, Hepworth, et al.) and in a sense by the ambi-
valent Porter. He was, moreover, quickly outdistanced by
Ince, De Mille and Barker. The Janus-like figure of Wil-
liam S. Porter is especially illuminating since although he is
in every essential respect a Primitive director, he is also one
of the 'inventors' of many of the modes of shot-change,
including the reverse field, cross-cutting and ellipsis, which
were to constitute the basis of the 'Hollywood codes.' 5 His
work very explicitly embodies the 'Lumiere-Edison con-
tradiction' which is at the heart of the dialectical develop-
ment of film during the first twenty-five years.
For example, Porter introduced a medium close-up into
The Great Train Robbery ( I903) in a pioneer attempt to
alleviate the 'impersonality' of those long shots in which the
characters never occupy full screen height and in which the
outlaws are distinguishable from the posse only in so far as
5· The term codes, insofar as it implies they wear the traditional bandana over their faces. How-
the existence of a totalizing semiological
system at work, will henceforth be used ever, he had this shot delivered to theatres in a separate
with inverted commas implied roll, so that the now celebrated image of a cowboy shooting
throughout. For although it is my firm
conviction that such a system- or set of into the camera (and what better way to acknowledge the
systems- has certainly been at work in
Western cinema at least since Griffith
implications of the spectators' forced involvement in the
and his contemporaries undertook what screen image) could be spliced on to either the beginning or
was. at least in part, a codification in the
strictest sense, the actual nature of these
the end of the film proper, a practice related both to the
codes, the hierarchical relations that 'open screening' practices of Primitive cinema6 and to such
organize them, the differing degrees of
permanence that distinguish them, the
modern experiments in the 'film-mobile' as Chelsea Girls.
threshold beyond which they give way Porter also introduced, in the final sequence of The Life of
to a mode of representation irreducible
to the language model, in short the
an American Fireman ( 1902 ), the r8oa reverse-angle set-up
analysis of these codes and their genesis on a grand scale. His camera filmed a fireman saving a
has not, despite the pioneering
prospecting of Christian Metz and
mother and her child from the second-storey bedroom of a
others, been formulated to my flaming house, first from the street and then from inside a
satisfaction. Jorge Dana and myself are
at present preparing a book on this
studio reconstruction of the room. Fearing, no doubt, that
matter. In the meantime, I continue to his audience would be confused if the shots were inter-cut
regard the 'codes' as a convenient bit of
shorthand.
in the manner that was to become common some years
later, he then proceeded to juxtapose the two complete
6. Repeated screenings of the same film
or group of films at a single sitting as
versions of the action, which was thus seen to unfold twice
already noted, but also spectators on the screen, in complete contradiction with the linear
seated on both sides of a translucent
screen, films run backwards for comic or
concept of narrative time which otherwise prevails
'magical' effect, mixtures of live and throughout the film, as it had throughout eighteenth- and
screen performance, etc.
nineteenth-century drama and literature and would con-
7· At one point in the subsequent tinue to do in American and most European cinema for
archival history of The Life of an
American Fireman (exactly when and
decades to come. 7
under what circumstances, I have not The gravest 'deficiency' of the films made between the
yet been able to ascertain. but this
version has been distributed by the
time of the Lumieres' first productions and those of Grif-
Museum of Modern Art) these two fith, judged by the criteria derived from the nineteenth-
shots were divided up and intercut so as
to 'reconstitute' temporal linearity (i.e.
century novel and drama, was the lack of individualization
'continuity'). of character, the absence of the persona. The 'people' on
the screen were mere silhouettes; not only did they lack the A PARENTHESIS ON
voices and the colour of the theatre to give them individual- FILM HISTORY
ized presence and 'human' three-dimensionality, but the
face, the only visual sign that could both distinguish be-
tween them severally and provide some insight into the
'interior' self, into the soul, was to all intents and purposes
absent from the screen, since characters were generally so
'far' and so 'small'.
The reasons for this were the same that prevented Por-
ter, in his historical experiment, from actually incorporat-
ing his medium close-up into the 'continuity' of his film, and
impelled him to let it wander about the perimeter, so to
speak. For it was still tacitly assumed, and perhaps actually
experienced, by all those who made film, that any violation,
not only of the single, frontal theatrical view-point, but of
theatrical distance, would lead to the breakdown of the
illusion of reality achieved in the single-shot 'full-frame'
scene, inadequate though that illusion may have been. The
audience was judged unable to relate, for example, subse-
quently projected details of a given tableau to the total
space defined by an earlier glimpse of that tableau as a
whole. Whether or not this conviction was based on actual
experience is irrelevant to this discussion. The first decade
of European and American film-making shows that this
was the basic assumption, just as the film-making of the
following ten years showed how 'correct' it had been; for it
took at least that long to establish the 'rules' of match-
cutting which were ultimately to overcome completely the
sense of hiatus, of disorientation which did at times result
from the first attempts by Griffith and others to juxtapose
several successive 'prosceniums'. 8 Contiguity matching was
introduced through the use of exits and entrances with
matching directions from shot to shot. As this procedure
developed it was ultimately established that two opposite 8. Exactly for whom it resulted is of
segments of pro-filmic space could be presented succes- course difficult to determine today. We
know that from our vantage point, such
sively in the same screen rectangle (reverse-field cut). And an impression seems inevitable, but it is
finally, through the use of eyeline matching, the spectator quite probable that audiences were less
'critical' then. What is certain is that
was made the mediator between two interlocutors (the some film-makers and, above all, the
mediator of their gazes and their imagined speech). first trade-paper critics (see Kauffmann,
Stanley, History of American
This eyeline match, the veritable keystone of the 'Holly- Film-Criticism) were preoccupied by
wood system', was the last piece to fall into place these imperfections (in this, the latter
were the vanguard of ideology,
(towards the end of the First World War), the last device to spokesmen for their class and not for an
become staple practice, several years after cross-cutting essentially proletarian public), and this
contributes much to an understanding
and even the ellipsis had become established figures. The of the relatively rapid 'progress' of
general adoption of the reverse field and of the 'correct' cinema, i.e. the progression of the
dominant system of representation in
eyeline match were part of the last, most crucial and most Europe and North America. In Japan,
difficult stage in the process of breaking down the barrier of as we shall see, with an audience and
film-makers less disturbed by such
'alienation' which, despite the ducking heads that purpor- 'faults' of continuity, etc., this 'progress'
tedly greeted the arrival of the Lumieres' train, informed was considerably delayed.
A FROZEN STREAM? the relationship between the Primitive film and its essen-
tially working-class audience. That process lasted twenty
years, from the first economic recession affecting the film
industry (c. 1907) and impelling the first search for a 'better
class' audience, intimately related to the first naturalization
in cinema of the middle-class norms of representation
(Films d'Art) and the second recession which led to the
introduction of synch sound, completing the project inau-
gurated by Smith, Porter, Griffith et al.- and imagined in
the early r89os by Thomas Alva Edison.
Needless to say, this is an extremely simplified over-view
of a complex and contradictory historical movement.
I wish to claim, however, that the phenomenon which
virtually all specialists of the Japanese cinema, Occidental
and Eastern, regard as the 'lag' between Japan and the
West prior to 1920 9 and even 1930, was actually the mani-
festation of a fundamental incompatibility between the
West's developing 'codes of illusionism' and Japanese
indifference to 'illusionism' in the Western sense. This
incompatibility determined, even through the 1920s when
Western influence was relatively strong, the preservation
within the Japanese cinema of traits common to the Primi-
tive cinema of the r89os, and even to some of the most
radical Western films of the 1960s and 1970s (as, for
instance, those by Warhol and Godard). They are trium-
phantly affirmed in the fully developed structures which
underlie the masterpieces of Ozu and Mizoguchi, Naruse
and Shimizu made during the 1930s and 1940s. And we find
them in the films of many other minor but 'uniquely Japan-
ese' directors. Before attempting to trace the development
g. Anderson and Richie assert that in
1913 Japanese cinema 'had not even left of this implicitly critical separation, we must locate its
the cradle'. Op. cit. p. 29. origin.

66
6. A Rule and its Ubiquity

We have already observed that it was in 1910 or 1911 that

the merger of the four existing production firms resulted in
the first Japanese major company. In 1912 it adopted the
name under which it is known today, Nikkatsu. This move
coincided almost exactly with a similar concentration of the
means of production and distribution in the United States
after which it was patterned, and with the world-wide
emergence of cinema as an economically profitable indus-
try. One now observes, as I have pointed out, a shift within
the dominant Western modes offilmic representation. The
so-called 'theatrical' period 1 is succeeded by that of codifi-
cation, to which Griffith (who began directing just one year
before the constitution of the first American film trust)
contributed so decisively. It is said that until that time
(1909-12) the Japanese adhered quite closely to the pattern
of development of Western film, submissive to what is
generally and categorically described as 'the influence of
the theatre'. This synchronism is confirmed by my experi-
ence, but is still open to question, as less than half a dozen
films of that early period of Japanese cinema (1897-1912)
have survived. This presumption of 'theatricality' also tal-
lies with the undeniable tastes of what was indeed a mass
audience. For although the Japanese cinema may have
been, during its first few years, an object of curiosity for the
upper classes/ it soon acquired a predominantly pro-
letarian public, a situation which prevailed the world over
for the next three decades. 3 In Japan, tastes in the matter of
performing arts had been formed by the popular theatre: I. Not in the narrow sense that the Films
kabuki until the mid-nineteenth century, its derivative d'Art and Famous-Players films were
'stagey' but in the sense that the whole
shimpa during the period that immediately preceded and of Primitive Cinema is regarded as
followed the introduction of cinema. The tastes of plebeian subordinated to theatrical forms (which
is of course neither simply nor
Western audiences had similarly been formed by vau- completely true).
deville, circus, magic lantern and other popular arts, but
2. Richie and Anderson, op. cit., p. 22.
these were viewed by the dominant bourgeois taste as
archaic forms, suitable at best for children and their nan- 3· In the West, the audience came to
include a fringe of the intelligentsia.
nies, for they were in complete contradiction with the fully This seems to have been true as well in
developed 'illusionism' of the dominant theatre of the Japan, but I have no actual data.
A FROZEN STREAM? period: Shaw and Ibsen, but also Feydeau and Sardou. The
integration of taste achieved in Tokugawa Japan and per-
petuated during the Meiji era and after, has never been
achieved in the West since the rise of capitalism, largely
because of the forms and intensity of class struggle. And it
is to the extent that these 'pre-bourgeois' forms contributed
to the constitution of popular taste at the turn of the cen-
tury that it may be legitimate to characterize certain aspects
of Primitive Western cinema as traces of an authentically
proletarian art. Conversely, its gradual contamination by
naturalism may be seen in part as a class strategy.
In kabuki and the related doll theatre, we are dealing, it
is true, with a typically bourgeois theatre: both matured
during the rise of the merchant class in the Edo period. In
them we see the manner and the extent to which the aesthe-
tics of the He ian period, which remained alive and domin-
ant well into the nineteenth century (despite a long co-
existence with Chinese canons of poetry and painting) were
'revised' to meet the needs of the emerging class. A cursory
examination of this development and of the nature of the
theatrical practices which issued from it is unavoidable
here.
Kabuki had origins as humble as those of the aristocratic
no -a prostitute-priestess dancing in a dry riverbed. Due to
the historical circumstances of its growth, 4 it remained a
truly popular art throughout the eighteenth and early
nineteenth centuries:
The rise of the kabuki theatre, as well as that of the coeval doll
theatre, was linked with the rise of the townsman, for it was the
chanin who principally attended the kabuki, and the kabuki
became the expression of the townsman's artistic tastes and ethi-
cal beliefs. The no theatre was familiar to the commoner, but the
kabuki was lustier stuff, more suited to the tastes of a newly
important economic class than the esoteric no. 5

The severe restrictions- theatres were often closed down,

plays banned, actors banished- placed upon kabuki by the
Tokugawa bakufu ('headquarters') were associated with
the fear which a ruling class (in this case the landed
samurai) entertains with regard to a rising class which
objectively threatens its power. Indeed, the economic
influence of the wealthiest chon in (townsmen) was already
very real by the mid-eighteenth century. The pattern is, in
this instance and in almost the same period, very close to
that of Europe. The ban on Beaumarchais' Mariage de
Figaro was a symptom of the contradictions that led to the
French Revolution.
4· See Ernst, Earle, op. cit., pp. r-23.
The history of the modes of representation attached to
5· ibid, p. 9· the rise of the bourgeoisie throughout the West, as well as

68
those later developed in connection with the rise of the A RULE AND ITS UBIQUITY
proletariat, amply demonstrate that an ascending class
requires more directly 'realistic' representations than the
dominant class. Japan was no exception to this rule. Kabuki
and even the doll theatre moved away, in several significant
respects, from the purely presentational theatre of the no,
towards (but only towards) a representational theatre.
Shimpa, which catered to the newly urbanized peasants of
the Meiji period, and shin-geki ('modern theatre'), directly
an expression of the struggle of the liberal bourgeoisie,
indirectly of the embryonic working-class movement, were
further steps. The no, we must remember, had been an
almost purely lyrical form, in which the narrative was
reduced to a commentary on some theme or conflict drawn
from a fund of folklore and old chronicles. The text itself
displayed the same polysemous and intertextual complex-
ity as Heian poetry, and was also intricately woven into a
sophisticated fabric of music and dance. Kabuki and the
doll theatre, on the other hand, had strong, amply
developed plot-lines, and though some of the material was
drawn from the semi-legendary past (often via the no
texts), many plays - and of course those of the great
Chikamatsu Monzaemon come immediately to mind- had
contemporary settings, drew upon actual incident and con-
stituted, to a remarkable extent, 'portraits' of the life-styles
of the chonin. They accurately described the economic and
social pressures of modern life and seemed, in general, to
follow the pattern of emerging bourgeois art in the West.
They were not, however, political in content, partly
because of the extreme vigilance of the censors. It quickly
became expedient, for example, when writing plays about
the celebrated incident of the forty-seven ronin (masterless
samurai) who avenged their lord's forced suicide by killing
his rival, a protege of the Shogun, to set it in the remote
Kamakura period, even though it had actually been a con-
temporary event. Moreover, as I have already pointed out,
there was in fact no revolution brewing, nor any middle-
class political activity whatsoever; the only manifest social
discontent came from the peasants, who rose up periodi-
cally to claim tax reductions and other limited reforms.
Despite this tendency towards bourgeois realism, both
the 'live' kabuki and the doll theatre nevertheless
remained essentially presentational arts and were still
significantly closer to the no than to any Western thea-
trical practice between that of the Elizabethans and the
modernist renewal of our century.

The terms presentational theatre and representational theatre

have been used to describe antithetical forms which the theatre

69
A FROZEN STREAM? may take. The kabuki, because of the nature of its means of
expression and the quality of its rapport with the audience, can
be called presentational. This is an abstract term, not referring
to a specific historical theatre, but useful in outlining a general
form, such as that which the theatre took in Greek civilization
of the fifth century, in the Elizabethan public playhouse, in the
contemporary theatre of Meyerhold, or in vaudeville. In the
presentational theatre, the actor does not lose his identity as an
actor. The audience does not regard him as a 'real' person but
as an actor acting. His make-up, costume, movement, and
speech emphasize the difference between the actor and the
concept of a 'real' person that exists in the mind of the audi-
ence .... The stage is distinguished from the rest of the theatre
building, but it is not conceived to be spatially discontinuous
from it. The actor, the audience, and the performance exist
within the same psychologically undifferentiated world ....
At the opposite pole from the presentational theatre is the
representational. This generalized form appeared in the Greek
theatre of the fourth and third centuries, in the European
medieval mystery plays, but it probably reached its ultimate
statement in the theatres of Antoine and Belasco at the turn of
the century. In the representational theatre every effort is made
to convince the audience that the stage is not a stage and that
the sctor is not an actor. To this end, the stage is disguised by
the use of settings, properties, and lighting so that it will appear
to be a specific and 'real' place. Various technical means are
employed to create in the audience a sense of spatial discon-
tinuity between the auditorium in which they sit and the stage
on which the play is being performed. In essence, the stage
becomes an area of illusion, while the auditorium remains a
part of actuality. The actor, although he may have to resort to
highly 'unreal' methods to do so, seeks to convince the audi-
ence by his make-up, costume, movement and speech that he is
a 'real' person, not an actor acting. 6

In contrast, kabuki's 'rapport with the audience' is of

special importance to us here; it exemplifies the Japanese
concept of reading incorporated in artistic/social practice.
The kabuki audience of the eighteenth and nineteenth
centuries generally expressed their appreciation by call-
ing out to the actors at the moment of their entrance
along the hanamichi (a ramp running through the
6. ibid., pp. IS-19. It should also be
auditorium to the stage and which of course is an essen-
pointed out that Ernst's excellent tial element of that 'presentational' quality described by
description parallels Brecht's Ernst) or while they were performing on the stage
distinction between epic and bourgeois
theatre. The fact that Brecht's name is proper. This tradition is preserved today by small groups
never once cited by the author of this
otherwise very useful study is indicative
of connoisseurs who occupy the front seats at every per-
of its ideological limitations. formance. These calls may be simply the (nick-) names of

70
the actors or else conventional exclamations, such as A RULE AND ITS UBIQUITY
'Matte imashita!' ('That's what we've been waiting for!'),
uttered just before or after a particularly well-known and
important passage in the play. These cries have nothing
in common with the bravos or outbursts of applause with
which connoisseurs of Western presentational arts, the
classical ballet and opera, momentarily blotting out the
musical text. express their delight at the execution of a
set piece which they have learned to admire for its diffi-
culty of execution. The reactions of the Japanese, uttered
in a vocal style which is both distinctive and closely
related to that of the performers themselves, are carefully
placed by those who utter them, either individually or in
close-packed clusters, in such a way as to become part of
the 'text' of the play, part of its musical rhythm and
timbre and its dramatic structure. The 'shouters' (in Edo
days, most of the audience) have an extremely intimate
relationship with that text, somewhere between that of
the audience for flamenco music and the music-student
with his nose in a pocket-score. Such is the 'complete-
ness' of the total effect, that one wonders to what extent
the codifiers of kabuki may have taken this factor into
account in conceiving the texture of the musical accom-
paniment, the pace and length of certain scenes: today it
is impossible to imagine the auditory space of kabuki
without these Webern-like flashes which actually seem to
organize the movement of certain episodes. 7 Audience
participation of this sort may be regarded as the acme of
presentationalism.
Kabuki, in rejecting representationalism, evolved a
type of performance which further bore the inscription of
its own production in the stylized 'femininity' of the
oyama (female impersonator; see Chapter 7) and in the
visible stage assistants dressed in black who remove
accessories when they are no longer needed, help with
on-stage costume changes, etc. Other strategies include:
the completely free contraction and dilation of narrative
time; the polysemy and intertextuality of the actual 'lib-
retto' (less intricate, perhaps, than in the tanka or haiku,
but still present); the rejection of illusionist depth in set
design and in blocking (both related to what remained
the fundamental attitudes of Japanese graphic art).
The popular wood-prints of the period, ukiyo-e, were,
it is true, a step in the direction of 'representationalism',
7· Curiously enough. no Western
since through them linear perspective was first popu- commentator, including the
larized in Japanese art. However, pictorial art of the perspicacious Ernst, seems to mention
this aspect of the 'calling' practice. I am
eighteenth century continued to differ profoundly from therefore unable to verify my
Western illusionist painting in the acknowledgement of hypotheses about the musical
codification of these cries or their
surface and of the frame-line as disruptive edge. It influence on the 'writing' of kabuki.

71
A FROZEN STREAM? continued, as well, to ignore 'centering' and its underlying
an thropocen trism. 8
Ihara Saikaku, who wrote novels and collections of
tales dealing specifically with the economic realities of his
day, at the same time remained faithful to a concept of
the signifier at work derived, as I have said, from Heian
poetry, via the haiku. Perhaps the most significant exam-
ple, however, of the resistance offered by Heian values to
representationalism appeared in the doll theatre. At the
time (1727) when a supreme degree of realism had been
achieved involving articulated fingers and eyeballs, as
well as texts which recounted the most recent faits divers,
a step was taken which made this theatre a paradigm of
the distancing effect and of the inscription of the process
of production. The three manipulators, as well as the
musicians, began to operate in full view of the audience,
which they had not done before! 9
It is important to consider the doll theatre, 10 especially
as its development hinged upon its association with one
of the musical forms of Japanese narrative art called
gidayu bushi (also incorporated into kabuki, especially in
association with the performance by live actors of plays
written for the dolls - an extraordinary instance of 'inter-
8. 'The earliest designers of this type of
textuality' in its own right) and which had as its direct
print ... from the first showed no descendant the art of the benshi, the live 'narrator' of all
compunction in handling the human
figure as they would any other element silent films in Japan, who was to play such a vital role in
of composition. Quite often this meant the preservation of the presentational character of Japan-
that a figure would be cut in half
vertically by one side or the other of the
ese cinema.
print with a resultant asymmetry which Here is the descriptive analysis of Roland Barthes, by far
is a marked trait of Japanese pattern.'
Hillier, J., The Japanese Print, p. 92.
the most perceptive observer of the doll theatre.

9. Donald Keene sees this as the final The Bunraku dolls are from three to six feet tall, little men or
manifestation of a 'law' governing the
entire development of the doll-theatre women whose limbs, hands and mouths can move; each doll is
and which called for an 'even balance' manipulated by three men in full view who surround, support and
between 'the real and the unreal'
(Keene, Donald, Bunraku, pp. I3-I4). accompany it. The master holds the upper part of the doll with his
This somewhat mechanistic right arm; his face is bare, 11 smooth, pale, impassive, cold as 'a
presentation of the phenomenon -
supported by a declaration made by freshly washed white onion' (Basho); the two assistants wear
Chikamatsu himself- is quite black and their faces are hidden behind cloth. One man, wearing
applicable, as far as it goes, to all the arts
of Edo. thumb less gloves, holds a large pair of stringed scissors with which
he operates the left arm and hand; the other, moving close to the
I o. Generally, though incorrectly,
called Bunraku, which is simply the floor, supports the body and walking feet. These men move about
name of the one Osaka troupe which in a shallow pit which leaves their bodies visible. The set is behind
preserved the art-form through the
period of its decline in the nineteenth them, as in a theatre. On one side, there is a platform for musicians
and twentieth centuries. and narrators; their role is to express the text (as one squeezes
I I. Actually, this is true only of certain juice from a piece of fruit). This text is half-spoken, half-sung;
plays, in particular those derived from punctuated with great sweeps of the plectrum by the shamisen
the no; the master is usually dressed like
the assistants. player, it is both measured out and hurled forth, with violence and
artifice. Sweating and motionless, these voice-carriers 12 are seated
I2. Pone-voix: megaphone, but used
here in its literal sense. behind little music stands on which is placed the 'master text',

72
whose vertical characters may be glimpsed from afar as the pages A RULE AND ITS UBIQUITY
are turned; a triangle of stiff cloth, clinging to the shoulders like
a kite, frames a face suffering all the tortures of the voice ...
The Bunraku thus practices three separate modes of writing,
and produces them to be read simultaneously on three different
planes: those of the marionette, the manipulator and the voci-
ferator. The effective gesture, the effecting gesture, the vocal
gesture. It is the voice that is basically at stake in modernity, that
particular substance of language striving to triumph in every
domain. The Bunraku, on the contrary, has a limited concept of
the voice, it does not do away with it but assigns it to a definite,
essentially trivial function. Indeed, the narrator's voice combines
extravagant declamation, tremolo, a feminine high-pitched tone,
broken intonations, sobs, bursts of rage, of plaintiveness, of sup-
plication and surprise, indecent pathos and indeed all the stock
recipes of emotion, openly developed on the level of that inner,
visceral body of which the larynx is the mediating muscle .
. . .Thus, without being eliminated (this would be a mode of
censure and would designate it as important), the voice is put to
one side (in the actual theatre the narrators occupy a side-
platform). The Bunraku gives it a counter-weight or rather a
counter-march in the form of gesture.
Gesture here has two faces: that of emotive gesture on the level
ofthe marionette (people cry over the doll-lover's suicide), that of
the transitive act at the level of the manipulators. In our theatre,
the actor pretends to act transitively, but his actions are never
more than gesture; on the stage there is nothing but theatre and
yet it is ashamed of its theatricality. The Bunraku, on the other
hand (and this is its defining quality), separates the act from the
gesture; it shows the gesture, reveals the act, sets forth both art
and work, reserves a mode of writing to each. The voice (which
can now be allowed with impunity into its most excessive reaches)
is flanked by a huge volume of silence in which other traits, other
modes of writing may be inscribed with all the more subtlety ....
All this, of course, is linked to the distance effect recommended
by Brecht. That distance is reputed impossible, useless or trivial in
our culture and hastily abandoned, even though Brecht very
precisely situated it at the centre of a revolutionary dramaturgy
(and the one no doubt explains the other). The Bunraku enables
us to understand how this distance functions: it functions by the
discontinuity of the codes, by this caesura imposed upon the
different traits of representation, so that the copy made on stage
should be, not destroyed but broken, so to speak, spared the
metonymical contagion of voice and gesture, of voice and soul in
which our actors are mired. 13

The unusual length of this quotation is justified by its dense

and multi-faceted relevance to all the principal themes
under consideration: the relationship of Japanese systems 13. Barthes, Roland, op. cit., pp. fiS.-75.

73
A FROZEN STREAM? of representation with regard to logocentrism, the irrele-
vance of the concept of originality in this culture, the divi-
sion of the representational process into distinctly separate
texts, relatable only through an act of reading, 14 and the
relevance of the whole to Western revolutionary thought
14. They are, in fact, no more 'naturally'
fused than the Chinese characters with and artistic practice.
the native syllabary, which This radical disjunction between signifiers in the doll
'communicate' with one another only
through an extremely sophisticated theatre is present in them5 and, to a lesser extent, in kabuki.
correlational reading, no matter how It made a profound, direct impact on the Japanese silent
'instantaneously' it may in practice
often take place. cinema and has indirectly affected the sound film as well.

74
7. Bulwarks of Tradition

Very shortly after films first began to be shown in Japan, it

became general practice to have a benshi, a live commen-
tator in the theatre to accompany the film with vocal expla-
nations. There is every reason to believe that this was not
based on a simple calculation that 'people aren't going to
understand'; it was a natural, i.e. culturally and historically
determined, development. It did not come about because
'the Japanese like to have things explained to them.' 1 The
assertion that 'the Japanese like to have the signifier dis-
joined' may be scarcely more convincing ; it does neverthe- I During the very first decade or so of
less point to the existence within Japan of a concept, a film-making in France, England,
America and elsewhere, when
'module' present in all human activity. screenings were held mainly in a
The benshi is not a bastard outgrowth of a specifically (presentational) amusement-park
context, this practice also held sway.
Japanese defect or 'convention' 2 as it is likely to be called The outside barker would often step
when it appears in the doll theatre and other approved inside to continue his spiel during the
screening as lecturer. In England and
cultural products. Neither is there anything intrinsically especially the U.S.A., another
'low' about the kabuki-derived genre called chambera or important source for this practice was
the 'lantern lecture'. As the mores of the
sword-play films 3 which, during the latter part of the silent bourgeois theatre took over, however,
era (192o-36), practically monopolized the screen. This the practice gradually disappeared; in
Japan it lasted until 1937. is still an
ideological repression of intertextual ramifications of two appurtenance of archive screenings. If
traditional arts and of the popular forms by which the early ever there was needed conclusive proof
of the profound otherness of the
cinema is rooted in them, is doubly significant. It reflects Western Primitive cinema by
the onus traditionally attached to the Western cinema comparison with the standard
post-Griffith product, this
owing to its plebeian ('theatrical') beginnings, the sense of extraordinary brief encounter with the
shame, not unrelated to the notion of Original Sin, which cinema of Japan provides it.
those beginnings continue to inspire. We know that these 2. This word is often used to mask our
theatrical beginnings are, in fact, the instrumentality of the ethnocentric repugnance for
non-transpctrent representation: it
camera stripped bare. Its essential transformational powers implies that in the culture which
(its production of meaning) are thereby acknowledged, produced the sign, it is read as fully
transparent, 'their representations are
since even filming the theatre stage as such (i.e. filming the just like ours', basically everyone is like
entire proscenium) destroys the representational effect, us, all have their conventions and all
luke them for 'the real thing'. It is just
causes the image to appear as that of a stage. The Lumieres' possible, however, that the audience at
attempts to give their production a certain entertainment the Buroraku aren't crying over
characters at all but over dolls, over the
status, as in La Mort de Marat ( 1897), were framed in such a 'convention' itself.
way that (an imaginary) proscenium-arch and frame-lines
3· 'Chambera' is an onomatopoeia
coincide: the de-personalization becomes more radical than suggesting the clash of swords (see Ch.
ever, since the characters are tiny puppets, overwhelmed II).

75
A FROZEN STREAM? by empty height. A scant decade later, when the need
for gentility began to assert itself in the vogue for
'serious' theatrical subjects, such films as L' Assassinat du
Due de Guise (1907) which were slightly behind their time
in terms of editing, nevertheless ignored the proscenium.
However, it was not until the system of narrative editing,
with its close-ups, matching devices, etc., was fully
developed that it became possible to recover the theatre's
power of characterization, personalization, etc. Paradoxi-
cally, it is for the development of this system that Griffith is
celebrated as the man who brought cinema out of the
theatrical stage! The so-called theatricality of the earliest
cinema is in many instances a rudimentary and fugitive
transmutation of the popular theatre of that period into an
objective prefiguration of modernist theatre and cinema.
It is not surprising that this period of cinema should
be, in Japan as in the West, the object of systematic
repression.
More generally, analysis of prevailing attitudes towards
the relations between the cinema and the popular theatre
of Japan reveal a phenomenon that I shall term 'repression
of the Japanese text'. Our presupposition is that Cinema is
One, just as Man is One, that the Hollywood codes are
those of Cinema, East and West, the Codes of Man! As a
consequence of this we also admit that an individual artist
who plays on them, disrupts or subverts them, may be a
genius- or a charlatan. Anyone who simply ignores them,
however, who pretends they do not exist and wishes to
preserve the cinema in the state which it knew before the
coming of the Codes, who keeps on 'doing his thing', is
either a fool or slightly backward. Yet this is exactly what
Melies and, to a large extent, Feuillade were doing between
1910 and 1915, until the former had to abandon the cinema
forever and the latter was condemned to making futile
efforts to 'catch up'. It is what the whole of Japanese
cinema did for some ten years after the start of the 'Griffith
revolution' but also, to a surprising extent, throughout the
1920s.
By Western standards, and in contrast with the situation
in Europe, the 1920s do not seem to have been a very 'rich'
period in Japanese cinema. It was this stubborn refusal to
'grow up' which nevertheless provided the conditions for
the remarkable preservation through the 1920s of several
of the basic elements of the 'primitive' attitude, and which
ultimately made possible the remarkable developments of
the 1930s. It is my contention that the benshi played an
historically positive role in this period of tacit resistance.
The function, perhaps the need, of the benshi derived
from the theatre of Edo and from the many solo narrative
genres which abound in Japan. In contrast with these by BULWARKS OF TRADITION
then respectable origins of their practice, the first genera-
tion of benshi seem to have been composed predominantly
of politically ambitious men intent on improving their orat-
ory and ex-street vendors seeking to rise in life. 4 Indeed,
the benshi soon became a public figure of considerable
importance; in the decade from about 1915, people went to
the pictures to hear their favourite benshi rather than to see
a particular movie star, and would call out his nickname at
the beginning of the performance in the manner of a kabuki
audience. 5 These men, who are generally felt to have been
of singularly modest intellectual capacities (at least until
around 1919), seem to have acquired a considerable say in
the actual production of films. If the finished work seemed
in some way unsuitable to their talents, they demanded
cuts, the shooting of new scenes; they wanted existing
scenes lengthened to allow for development of their dis-
course (e.g. touching farewells). Above all, they fought
bitterly against the introduction of new narrative structures
such as the flashback. We will examine the consequences of
this attitude in the next section. It may already be evident,
however, that the conservatism with which the benshi is
commonly taxed primarily reflects his efforts to save not
only his means of livelihood but also a certain mode of
presentation which was the central artistic ideology of his
class and nation. We can, on the evidence of this limited
data, concur with those who see the influence of the benshi 4 This and other information in this
section, I owe to private conversations
as profound and diverse, without accepting the all but with the Japanese film scholar Yoshida
unanimous opinion that his role was harmful, that he Tieo, who has published in Japanese a
history of the benshi ('Katsuban no
'retarded' the growth of the Japanese cinema.6 rekishi' in Eigashi Kenkyun, nos. r & 2,
Needless to say, the films of this earliest period had 1973). The rest ofthe information is
from Richie and Anderson unless
absolutely no inter-titles (or 'spoken titles', as they were to otherwise stated.
be called in Japan when they were first introduced in the
5. There is no evidence as to whether
early 1920s), except for an occasional indication of the this was also done during the film; nor
setting to follow or a title given to a section of the film. 7 In do we know how, if at aU, it was related
to the benshi's delivery or to the music
the early days (roughly, until about 1912), the benshi not that accompanied him.
only supplied a voice for all the characters but provided a
6. The title of Chapter 2 of a recent,
running commentary on every detail of the image and rather sketchy history of Japanese
action, often repeating himself in chanting patterns if he cinema expresses this attitude: 'Exit
Benshi, Enter Beauty'.
ran out of anything new to say. The style of many benshi
was fairly straightforward, but variously shaded by the 7. Deslandes and Richard assert
(Histoire comparee du cinema, Vol. 2, p.
techniques of other narrative arts. And of course there was 201) that 'titles do not seem to have
a musical accompaniment, consisting generally of a mixture come into general use until around
1901.' This would be another link
of native Japanese instruments (shamisen, taiko, etc.) and between the earliest primitive cinema in
such convenient European instruments as the violin. 8 It the West and the Japanese films of the
first twenty-five years.
seems quite certain that the overall acoustical effect had
little of the plasticity ofgidayu bushi. At the same time, and 8. Eventually the musicians were
supplied with 'repertoires', as in the
though admirers of the traditional theatre may only scoff at West: scraps of music corresponding to
this degraded version of a sophisticated art, the effect this or that mood.

77
A FROZEN STREAM? described by Barthes as a fragmentation of the representa-
tional gesture could not help but be produced.
It would be naively ethnocentric of us to consider the
institution of the benshi as simply 'a crude dubbing effect', a
puerile attempt to make the picture talk before the advent
of sound. The time did come (in the 1920s) when for Japan-
ese (as opposed to foreign) films, several such speakers
divided up the various roles. Even then, however, the
desire for ever greater realism can scarcely be regarded as a
primary cause. Such a desire was infinitely less strong
among the Japanese of the first quarter of the century than
in the Western middle class- for of course it was they who
shunned the silent cinema and came in droves to the talkies,
thereby determining the second and final stage of the pro-
cess by which Western cinema was at last fully subordi-
nated to a middle-class ideology of representation.
The key to these 'film performances' 9 is the fact that
through the benshi the image was purged of speech and
relieved to an almost equal degree of the narrative burden.
In a sense, the Japanese silent film was the most silent of all,
if by silence we mean, as most people do when they talk
about that film era, the absence of speech. Speech was
indeed explicitly absent, since it was removed, put to one
side; the voice was there, but detached from the images
themselves, images in which the actors were thereby all the
more mute and were confined, moreover, in many
instances, to remarkably static visual renderings of the
scenes unfolding through the voice, much like the dolls or,
to a lesser extent, the kabuki actors of the Edo stage. 1" And
9· Of course, they weren't really film
performances at all, since often the
the 'transference' of the written word from its Western
benshi was felt to be the centre of interest position between the pictures to a 'libretto' on the benshi's
(though in practice it is probable that
there was an oscillation of that centre
lectern is as significant here as the analogous phenomenon
between benshi and screen as happens in the doll theatre. In the dominant cinema of the West, the
in the doll theatre). It is this
'de-centering' of the performance that
dialogue titles always made it clear that Speech, the Word,
most outrages the 'serious' filmgoer was an intangible, ineradicable presence inside the diegesis,
today, with his quasi-religious attitude
towards the screen (exemplified by the
that printing was merely its passive outward vehicle: this
pew-like seating architecture of the first was implicit in the way in which the title demanded a
Anthology Film Archive Theatre in
New York). momentary suspension of the images. It was isolated in a
decorative frame, and against a timeless black background,
10. A film made in 1922, at the height of
the benshi's popularity, by one Oboro
signifying the parenthetic suspension (not the ac-
Gengo, Two People Named Shizuka knowledgement) of representation. Except in very rare
( Ninin Shizuka), consists largely of
static conversations between seated
experiments, such as L'Herbier's L' Homme du large, it was
figures and contains absolutely no never super-inscribed on the picture, as it so often was in
inter-titles other than
'chapter-headings'.
the Japanese films of the 1920s, a phenomenon we shall
examine in the next section.
r 1. Contrary to the practice among
Western nations of supplymg untitled
To return for a second to the 'libretto' just mentioned, it
prints and letting distributors title them should be explained that it actually came to exist though for
in the local tongue. Japan was supplied
with exhibition prints prepared for the
foreign films only, since they, contrary to the Japanese
domestic use of the exporting countty. product, did have inter-titles ... in their native language. 11
The benshi was nominally expected to translate those titles BULWARKS OF TRADITION
in addition to giving his usual comments on the heroine's
dress, the weather, etc., but in actuality he often made up
the lines entirely, and significantly enough, even changed
the narrative meaning of the images at his pleasure - or
rather in accordance with a cultural outlook shared with his
audience. It is also said that the characters in foreign films
were nearly always given the same names: Mary for the
heroine, Jim for the hero and Robert for the villain. 12 This
practice may be seen as a de-construction of the Hollywood
film, read in terms of its stereotypical structures by a cul-
ture which values the stereotype. We may, in fact, consider
the benshi's entire discourse as a reading of the diegesis
which was thereby designated as such and which thereby
ceased to function as diegesis and became what it had in
fact never ceased to be, a field of signs. The most 'transpar-
ently' representational film, whether Western or Japanese,
could not be read as transparent by Japanese spectators,
because it was already being read as such before them, and
had irrevocably lost its pristine transparency.
The benshi removed the narrative burden from the
images and eradicated even the possibility of the images
producing a univalent, homogeneous diegetic effect. In the
West, on the other hand, the need for such unity was so
strongly felt that it gradually resulted, towards the end of
the silent period, in a tendency to do away with titles
altogether or nearly so, and 'let the pictures tell the story'.
Films like Der Letzte Mann, Menilmontant, Ueberfall
attempted to refine existing codes of decoupage to a point
where it became possible to dispense with the disrupting
titles and yet maintain the control of the flow of significa-
tion, the linearity of the narrative, the 'naturalness' of the
diegesis. Perhaps the most systematic and sophisticated of
these attempts was that by Kirsanoff in Menilmontant.
However, the fundamental semantic ambiguity of the
uncaptioned photographic image confers upon this film an
obscurity which may be responsible for the interest it has
aroused on the apparent assumption that failed prose is
poetryY This tendency towards the 'all-picture movie' had
a curious development in the early 1930s, especially in the
United States. Many prominent directors, out of an instinc-
tive distrust of canned theatre, felt that 'the picture should
still tell the story', as they might have put it, and often went
to extravagant lengths (at times with preciously witty 12. Richie and Anderson, op. cit., p. 25.
results) to develop a point of narrative without relying on
13. See, in particular, Michel, WalterS.,
titles or dialogue, using sound only as a 'counterpoint' (an 'In Memoriam of Dimitri Kirsanov, a
elaborate dolly-shot linking two hotel balconies in order to Neglected Master', in Sitney, P. Adams
(ed.), Film Culture Reader, Praeger,
co-locate two key sets, a series of clock-face close-ups and New York, 1970; Seeker & Warburg,
snatches of off-screen dialogue to indicate the passing of an London, 1971.

79
A FROZEN STREAM? evening are two typical examples from Lubitsch's Trouble
in Paradise of 1932). Interestingly enough, these develop-
ments were not unrelated to Eisenstein's aspiration to an
'intellectual montage'- the expression of a relatively com-
plex theoretical discourse through images alone, whose
degraded equivalent is the 'montage sequence' of domin-
ant cinema - which crystallized at about the same time
(1927-30). It is very significant that Iwamoto Kenji, 14
asserting that the institution of the benshi amounted to a
rejection of the narrative codes of editing, cites not Griffith
first and foremost, but Eisenstein. And indeed for a cinema
indifferent to the organization of images in accordance with
the language model, the 'Griffith' codes and Eisenstein's
montage of attractions and intellectual montage can be
considered equivalent. This matter will be developed in
connection with the cinema of the 1920s.
In conjunction with the semantic dissociation effected by
the benshi, it is necessary to mention a material dissociation
which, although less extensively practised, is nonetheless
highly symptomatic. It appears that in the earliest years of
film-showing, at least one exhibitor set up rows of seats
allowing spectators who wished to do so to watch the projec-
tion rather than the film. 15 Together with the early, quite
general practice of commencing every performance by a
demonstration of the workings of the projector, this would
seem to provide evidence that the cinema made its debut in
Japan under the auspices of what we may well call, again
taking our cue from Barthes, the co-presentation of the
effective and effected gestures.
By 1915, practices of this kind were anomalous in the
cinema of the West. In the earliest period they were not. As
we know, the projector was in the hall, shop or tent for
several years, and we have already mentioned the early
absence of titles and the ephemeral Western equivalent of
the benshi, the lecturer. It is also interesting to note that the
pronounced flicker effect, whose final eradication in 1909
(through the invention of a new type of projector shutter)
significantly coincided with the beginning of a veritable
mutation in the modes of representation of Western
14. Unpublished essay in Japanese. See
cinema, was presumably present to a far more 'irritating'
note 8, page 98. extent in all of Japanese cinema until around 1912 and in
15. Richie and Anderson allude to this
much of it well into the 1920s. For it was the practice in
( op. cit., p. 24) and I owe some Japan to shoot at twelves frames per second, 16 rather than
complementary data to a private
conversation with Mr Richie. The
the sixteen io twenty frame average which was customary
original source is Tanaka Junichiro, op. in the West. In Japan, as in the West, the projector, like the
cit.
camera. was hand operated; it is probable that projection-
16. According to Sa to Tadao, some ists varied the speed of the film as they did in the West,
(oral) sources speak of eight frames per
second, but this practice was unlikely to
according to the nature of the action. Needless to say, the
have been very common. flicker produced at such low speeds was one which had

So
rapidly become intolerable in the West, impinging as it did BULWARKS OF TRADITION
upon the illusionist force of the filmic image, if only
through the eye-strain induced by the viewing of films that
were increasingly long.
The 'look' of the Japanese film during the period of the
Western Primitive cinema was very close to that of the
European film. It is with the beginning of the Formative
period that differences begin to appear, 17 since the Japan-
ese continued along the same path. Japanese film-makers
were quite aware of new developments in the Western
cinema; Western films were, after all, shown in Japan with
increasing frequency and, according to Richie and Ander-
son, 18 Griffith's pioneering short films had already been
seen there by I9I3. The common assumption is that the
Japanese directors paid no attention to these innovations
from abroad and did not, in consequence, know how to use
them. We have evidence that this was not the case. In this
matter, too, we encounter the ethnocentrism of Western
scholars, and we must reckon, as well, with the unfortunate
tendency among all but the youngest Japanese scholars to
accept unquestioningly Western and 'Western-type' ideas
and criteria regarding the cinema. They are unconsciously
inclined, one feels, to consider films as the 'natural prop-
erty' of the West.
If we examine what may or may not be Makino's first
full-length version of Chushingura (A Tale of Loyal Retain-
ers, commonly called 'The Forty-seven Ronin'- I9I3 or
I9I7) 19 we find that he 'kept his camera running without
interruption through an entire sequence and never moved
it from its front-on angle of a spectator at a stage-play', (see
Fig. I) and 'completely ignored Griffith's editing con- 17. The differences could be observed in
Japan, not in the West. The first
cepts.'20 Except that he did not ignore them completely. Japanese film to be shown in the West
There are, in fact, several match-cuts in this film ( concer- (to non-Japanese audiences, at least)
was Kinugasa's Crossways ( 1928).
tinas) which show that Makino had mastered the technique
of cutting on movement far better than Louis Feuillade 18. Op. cit., p. 32.
ever did, far better, indeed, than most Western directors 19. There is a serious conflict as to the
did until about I9I5. In any case, we need hardly dwell on date of this film: Richie and Anderson
give it as 1913, the catalogue of the
the well-known rapidity with which the Japanese grasp new Matsuda Film Library, where it is
techniques. There are also lateral re-framing pans which preserved, as 1917. I find the later date
better substantiated by internal
show that his cameraman was skilled in manipulating evidence, but in the event that the
tripods that were every bit as manoeuvrable as those of earlier were proven correct it would
certainly give even greater weight to the
Hollywood (they might even have been American or claim made here.
French). 21
20. Richie and Anderson, op. cit., p. 32.
Most significant of all is the brief but dramatically capital
sequence in which Lord Asano attacks Kira, an act for 21. The earliest 'theatrical' production
that I have seen, Chronicle ofTaiko (c.
which he will be sentenced to commit ritual suicide. This 1go8, director unknown), also contains
tableau, the starting point of the whole saga of the Forty- several perfectly smooth lateral and
vertical pans serving to enlarge the
seven, is immediately followed, during the moment of fragment of a kabuki set used as
'shocked reaction', by a pair of medium close-ups in background.

8I
Fig. 1. Makino Show, A Tale reverse-field construction,22 bracketed by a pair of concer-
of Loyal Retainers (1913 or tinas. This passage occurs very near the beginning of the
1917) film, and with the exception of a few more deftly executed
concertinas, such editing never recurs. 23
As later developments amply demonstrate, the Western
codes had impinged upon Japanese perception, but Japan
was on the whole not interested in them as a system; they
were merely used on occasion to produce special dramatic
effects. The implications of this are clear enough. By 1913,
22. I use the term reverse field in this certainly by 1917, the reverse field and concertina, the
historical context to designate its most
general form: a contiguity cut involving
medium close-up, even the cut on movement, were in the
eyeline matching. It was not until the United States and most of Europe, banal devices. When
early 192os in the West that the actual
camera reversal, with eyelines close to successfully executed, as they increasingly were, they were
lens-centre, was added to the illusionist not perceived as the signifiers of anything more strictly
edifice, ultimately to become a keystone
of that edifice.
defineable than 'continuity', 'contiguity' and other basic
semes of linear representation. Makino, in contrast, uses
23. Even if the film is a compilation,
such as were often made and shown just them as privileged dramatic signifiers, comparable to the
after the Second World War, at one signs used in Japanese theatre to displace the gestures of
point in his early career, Makino did
film this long sequence consisting of two
emotion (the oyama tugging with her teeth at the sleeve of
typically 'primitive' shots, interrupted her kimono to signify weeping). This reading of the Makino
by this brief 'flurry' of editing. The
perfection of the matching (movement, film would scarcely be conclusive were it not for the fact
costnmes, lighting, make-up, sets, that throughout the 1920s and particularly during the late
emulsion quality) excludes the
possibility that the close shots were years, for reasons which I shall attempt to make clear, the
added after the fact. Western mode of decoupage was to be greeted in Japan
with three distinct attitudes. The first of these was utter BULWARKS OF TRADITION
unconcern, underlined by the rare but 'technically correct'
introduction of Western editing devices either as privileged
signifiers, as in the early instance just cited, or in a way
which can only be described as random. The second
response was the adaptation of devices as signifiers of a
completely different and more 'open' code; the signifiers
thus adopted were secondary, specialized signifiers in their
Western context, such as the swish-pan or dissolve. The
rarest response was mastery and constant utilization of the
codes according to the norms of Western practice; this
seems to have been true of Ozu and Mizoguchi, whose
earliest films are, however, presumably lost. It was, as well,
the case of one Futagawa Montabe, some of whose films
have come down to us and who, by 1928, had nothing left to
learn from Hollywood. We will explore the interplay of
these conflicting currents in the next section.
The overwhelming majority of films from this early
period, then, drew their substance from the kabuki reper-
toire, or from the repertoires of its derivatives, shimpa and
shin-kabuki (modernized kabuki). Scholars dispute the
importance of kabuki's 'influence' on the Japanese
cinema. 24 This is due to a misunderstanding about the very
nature of influences. I shall attempt to clarify this matter in
the next section, in attempting to reconstruct the historical 24. D. Richie feels that it was
insignificant, while Iwamoto Kenji
role of the-benshi during the twenties and in defining the holds the opposite view. This is a
most general relationships of the whole of Japanese theatre corollary of the degree of importance
which they attach respectively to the
to the cinema. In this early period, however, there can be popular cinema of the 19205 and before.
absolutely no doubt that the pertinent visual traits of
25. Until the advent, that is, in 1go6 of
kabuki appear constantly on the screen as a surrogate of shin-geki, the first 'reasonable facsimile'
that presentational character defined by Earle Ernst as in Japan of Western theatre, and which
was to play its role in the cinema when
common to all Japanese theatre. 25 They helped to preserve the Western codes were introduced in
the Japanese cinema against the ideology of 'realism' which the twenties (see Chapter 10).
rapidly took over the cinema of the West. The stylized 26. It is interesting to compare this
fighting sequences in which no actual blows are exchanged, kabuki-derived practice with a similar
device used by Griffith in A Corner in
the use of the backward somersault signifying the death of a Wheat (1909). The characters are shown
fighter, the translation into Melies-like 'special effects' of in a frozen tableau vivant between two
black-outs, a device clearly derived
the transformational machinery of kabuki and, above all, from the vaudeville and melodrama
the action-stopping mie, or tableau vivant (which termi- stages where it had, as in this film, a
function diametrically opposed to that
nates most of the scenes of Makino's early Chftshingura), 26 of the mie: it introduced into a 'life-like'
are clear 'distancing devices' which need no special elucida- context a 'symbolic', 'allegorical'
gesture designed to generalize the
tion. Less self-evident, perhaps, is the even greater impor- meaning of the play/film. It is true
tance of the use in film of traditional theatrical make-up enough that on the stage as in the film,
such suspensions were at odds with the
and its corollary, the oyama, or female impersonator. dominant need for continuity and
It is not surprising that one of the great 'battles' fought homogeneity, and were soon eliminated
from the theatre and cinema of the
just after the First World War, at the beginning of the bourgeoisie. Ambiguous as they are,
movement to import the Hollywood codes, was over the however, they are far removed from the
mie, which functions as the quintessence
elimination of the oyama from films. Her presence was of the presentational attitude and has no
absolutely inimical to those codes whose goal is to be 'expressive' dimension whatsoever.
A FROZEN STREAM? schematically summarized in terms of the 'psychological
depth' of the image and the fusion of the diegesis with a
'real world' in which women are women and men are men.
As for the theatrical make-up which remained customary in
chambera until the mid-1920s, it was of course in complete
contrast with that of the West. Although still often visible
as such until about the same period, make-up in Western
film was designed to heighten the expressiveness of the face,
felt to be diminished by the absence of words, of three-
dimensionality and of colour. Kabuki make-up, on the
other hand, is purely graphic; it reduces, on the contrary,
both the expressiveness and the singularity of the face.
One of the traits that strikes the modern viewer as most
significant in Makino's early Chushingura, is the fact that
the super-star, Matsunosuke, plays at least three different
roles. He is first seen as Lord Asano. After the latter's
forced suicide, he acts Oishi, the Lord's principal retainer
and organizer of the ronins' vendetta against Kira. Finally,
during the vendetta he plays Kira's principal bodyguard.
This practice also derives from kabuki, in which certain
plays call explicitly for double roles, and spectacular
stage-business is derived from the possibilities thus
created. 27 Its preservation in films, especially without any
change of make-up (as was the case, not only in Makino's
early Chushingura but, I am told, in many films up through
the mid-1920s), implied, of course, the audience's perfect
familiarity with a story like Chushingura. It also required
the presence of the benshi in the theatre to identify the
successive characters for any members of the audience
whose memory might be deficient. The benshi's task was
not to 'restore' to the image some virtual reality, garbled by an
unfortunate effect of the star system, but simply to name the
27. The heroine and one of the male roles which were, quite unmistakably, avowedly being played
principals in Tokaido Yotsuya Kaidan
(Ghost Story ofYotsuya) (1825) are
by one and the same actor, by Matsunosuke. Once again the
played by the same actor, which allows benshi was part of a fragmentation of the signifier. 28
for the presentation in rapid succession
oftheir dead bodies on opposite sides of We find the most remarkable instance of disjunction in
a door floating down a river in one of the early Japanese film in a genre known as rensa-geki (chain
play's most famous scenes. (It is seen in
Mizoguchi's masterpiece, Tale of Late theatre) which presented shimpa pieces whose interior
Chrysanthemums). scenes were performed by live actors on stage whilst
28. Of course, the hypothesis that this is
exterior scenes were performed by the same actors on a
a composite film might modify the motion-picture screen. Though the filmed sections of one
reading of this practice here. However, I
have been assured (by Sa to Tadao) that
or two of these pieces have survived (in the Matsuda collec-
this was indeed a general practice at the tion) we have only scant accounts of the actual nature of the
time. Moreover, audiences did at one
time or another see and presumably
full performances. The most remarkable quality of rensa-
'accept' the feature-length film which I geki, however, was its long life (1904-22). Similar attempts
have seen. And the relationship
between the images and the discourse of
at mixed media performances were made in the West,
the benshi (or naniwa bushi, a form of within the period we call Primitive (here again the
narration added to silent films for early
post-war revivals of such films) was encounter with the contemporary avant-garde is striking).
inevitably as I describe it. An acrobat from Jean Durand's troupe, the Pouittes,
opened a roof trap-door (on film) and clambered down a BULWARKS OF TRADITION
ladder on to the actual stage of the Gaumont Palace. In
Australia, one of the world's earliest feature-length audio-
visual productions (staged, c. 1906, by the Salvation Army)
was a mixture of film, lantern-slide and scenic tableau. All
such Western experiments were, however, extremely
short-lived. 29 Their discontinuous use of different media
was incompatible with the unity of the 'illusionist' system.
In this connection, one should also cite the juxtaposition,
in the early version of Chushingura already discussed and
in many other films up to the late 1920s, of tableaux repres-
enting out-of-door scenes by ostentatiously painted back-
drops, with other scenes actually shot on location. These
latter, moreover, would often involve the papier-mache
props of the theatre. Examples of a related practice are to
be seen in the remarkable semi-documentaries produced
by Pathe around 1906- Le Mineur, Sauvetage en mer, etc; by
the end of the war, however, such a practice had been
banished from the Western cinema, since it had quickly
become a hindrance to credibility.
It is but one step from this evocation of the use of
back-drops in the early films to the vital problem of the
representation of space as practised in Chushingura and in
most other films of the period. This is a matter of far-
reaching implications; it raises the problem of illusionist
depth, in both a literal and a figurative sense of the word.
For the moment, it will suffice to observe that the flatness
derived from the kabuki and doll theatre stages seems to
have been the general rule in the early films; 30 it is a trait
which lasted late into the 1920s, and left an indelible mark
on the films of some of the masters of the 1930s. The same is 29. Earle Ernst's erroneous description
true of the most striking compositional feature of these ofrensa-geki as a 'nine-day wonder' ( op.
cit., p. 252) indicates our difficulty in
early films, also related to the flattening of the image: there conceiving a 'bastard' form such as this
is frequently an inordinate amount of empty space over the as achieving real mass popularity within
the context of cinema as our society has
actors' heads, owing to the simple fact that most of the time shown us it 'must' be.
they are sitting on the floor. No effort is made to compen-
30. A Makino film entitled Thunder Boy
sate for this, whereas it would have automatically been (Jirai-ya), a good example of the
made in 'advanced' Western films from about 1914 on. This Melies-like trick films which were a
speciality of his, involves a considerable
de-centering, which was one of the principal traits of Primi- amount of axial movement in location
tive cinema in the West - where it also contributed to a shots, which shows that he, at least, was
not insensitive to this lesson of Western
flattening of the image, since the background appears to be cinema. However, the date of 1914
above rather than behind the figures- was preserved in the assigned to this film in the Matsuda
catalogue seems very unlikely (in many
Japanese cinema for decades to come and may still be other ways, the film seems more
perceived to this day. In fact, there are grounds for main- Westernized than the early
Chf<shingura) so that it is difficult to
taining that just as Japan has acted as the custodian of much know how meaningful this example was.
ancient Chinese culture that has been lost in the vicissitudes In any case, axial movement did not
become common in Japanese films, so
of her great neighbour's history, she is also the 'storehouse' far as I have been able to ascertain, until
of what were universally the Primitive modes of filmic the late 1920s, one of the earliest
examples I know being Makino's last
representation. version of Chf<shingura (1928).

ss
A FROZEN STREAM? It is also possible to detect, during this early period, signs
of the problem elicited by the 'realist vocation' of the
camera, i.e. its faculty for recording facsimile representa-
tions of perceptual reality. The pressure exerted by the
'Western-ness' of the machine is implicit of course in the
use made of the camera in rensa-geki. It is perhaps more
significantly to be observed in the subtle changes brought
about in the techniques of kabuki as they were transferred
to the screen. One such example is to be seen in the very
early film Chronicle ofTaiko (Taiko junanme, c. 1908): it is
part of kabuki practice that whenever an important prop-
erty has ceased to serve an active purpose on stage it is
removed by one of the black-clad stage assistants
(kurombo ). In Chronicle of Taiko, the hero at one point
lays his bowl-shaped straw hat on the ground. In a live
performance, it would have been deftly carried off by an
'invisible' stage assistant; here, however, it is whisked out
of shot at the end of an invisible thread. I have, in fact,
never seen a silent film in which the stage assistants appear,
however close the film may have seemed to theatrical prac-
tice otherwise. Considering the degrees of stage artifice
which became 'naturalized' in the Japanese cinema for so
long, it is difficult to understand why this particular practice
seems to have been systematically excluded, unless it can
be seen as one of the first effects of Western attitudes
defining the cinema as more 'realistic' than the stage, and
thereby as the first sign of a conflict which was to persist
throughout the next two decades and which was to deter-
mine the entire course of Japanese cinema. It is to the
development of this conflict that we must now turn.

86
Part 3 Cross-Currents
8. Transformational Modules

There is an awkward problem which the observer of

things Japanese must confront. It is one to which we
have already alluded in its ideological formulation: the
uniquely Japanese faculty for assimilating and transform-
ing elements 'borrowed' from foreign cultures. To my
knowledge, no substantial effort has yet been made in
the West to define or analyse this phenomenon, though it
has often been commented upon.
We are dealing with Japan's use of a foreign machine,
with a medium which has been largely dominated by
Western capitalism. These two circumstances could not
fail to have some specific impact on the cinema of a
nation whose goal has been, for over a century, that of
taking her place among the 'Powers'.
This historical pattern, this 'borrowing', was not, of
course, limited to Japan's relations with her Western
'mentors'. Long before Europe and the United States
came to be viewed in this role, she was China-oriented.
It has been suggested that an inventory of the perti-
nent traits of Japanese social and artistic practices would
seem to designate them as radicalizations of traits charac-
teristic of one or more of the great mainland cultures
(China, India, Tibet). This assertion must be understood
in terms of the political and ideological co-ordinates
within which it is made. For not only does Japan offer
traits which seem even more remote from our own,
Western ways of thinking and doing, more remote than
comparable traits of other Far-Eastern societies; these
traits also lend themselves to a Marxist critique of mod-
ern Western history in many of its aspects. The claim for
the radicalism of Japanese practice is therefore not to be
taken as the assertion of some metaphysical absolute, but
as an attempt to describe the situation which Japanese
'culture' effectively occupies today with regard both to
the dominant ideological profile of Western Europe and
the Americas, and to those practices, scientific, literary
and artistic, which instantiate the Marxist critique of that
dominance.

8g
CROSS-CURRENTS The key to an understanding of this process of radical-
ization in Japan might be suggested by an inventory of
the transformational modules that can be observed during
the periods of Chinese influence and of subsequent Wes-
ternization. These modules bear directly upon the ways
in which the codes developing in the cinema of the West
between 1900 and 1920 were transformed, displaced,
truncated in Japan during the 1920s and 1930s.
To begin with, we may say that the Japanese have
reacted to foreign importations - ideas, techniques and
artefacts - by wholesale acceptance, global rejection or
transformation/adaptation. Obvious as they may seem,
their unparalleled importance in Japanese history sug-
gests that the singular circulation of these modes of reac-
tion, the remarkable constancy of their social integration,
is of special interest to us. These reactions, moreover,
often co-existed in overtly contradictory forms, a factor
of importance for our discussion of cinema proper.
Foreign methods and, more often, techniques have
been adopted in whole insofar as they proved compatible
with the geo-political context of Japanese reality, and, in
particular, with the goals of the ruling class; hence the
rejection of the administrative structures engendered by
the Asiatic Mode of Production and imported for a time
from China but which proved totally unsuited to Japan.
In areas of artistic practice, the principle of wholesale
adoption is often applied in an apparently contradictory
manner. Foreign procedures which are fundamentally at
odds with native methods are nonetheless adopted in toto
and thrive alongside their contraries, as in Chinese
Buddhist architecture and its emphasis on ornate detail
and emphatic contours, antithetical to the geometrical
simplicity of native architecture. The busy, furniture-
crowded room of Meiji times occupies something of the
same position with regard to Japan's Westernization as
the neo-Chinese pagoda to her sinicization; both are
emblems of cultural interaction at the level of practical
social procedures. So were the emulation of Western sci-
ence and industry, the temporary adoption of Chinese
administrative procedures, the permanent adoption of
Chinese writing and Confucianism. It should be added
that sudden, wholesale adoption has always been linked
with the need of a ruling or a rising class to reject 'tradi-
tional values' in order to increase or perpetuate its
power. For indeed, the disavowal of such values may
alternatively favour, according to circumstances, either
the perpetuation of power under a new guise (the imper-
ial family in pre-feudal times, the feudal oligarchy in the
middle of the nineteenth century) or the rise of classes
excluded from it (the liberal middle-classes after the First TRANSFORMATIONAL
and Second World Wars). MODULES
It is, of course, the median category, the process of
adaptation and transformation, which has been most pro-
fusely illustrated throughout Japanese history. I shall cite
only three symptomatic instances: the ritualization of
Indo-Chinese Buddhist logic, already described, the
introduction of linear perspective during the period of
Western influence which immediately preceded the open-
ing of the country in 1852, and the adaptations of West-
em clothing which were so prevalent during the Meiji
era.
Nakamura Hajime 1 regards the rejection of 'logic'
('illogicality') as a basic trait of the 'Japanese way of
thinking'. Schooled in Occidental logic, this eminent
philosopher understandably regards this trait as a funda-
mental deficiency. However, it is also clear that in his
view the cultural filter which brought about the final
ritualization of Indian logic is closely related to the
Japanese rejection of metaphysics, to their emphasis on
the material world, on the 'here and now', on the 'limited
social nexus'. And that all of these factors bear upon the
ultimately 'non-transcendental' character of all religious
practice in Japan. 2
Specifically Japanese modes of pictorial representation
(which may be traced from the early hand-scrolls of the
Heian period, through the screens of the Muromachi and
early Edo periods down to the ukiyo-e prints of middle
Edo) exclude the concept of depth representation more
radically than Chinese painting, which made very pre-
cocious and systematic use of aerial perspective and
sophisticated compositional techniques suggestive, albeit
ambiguously, of illusionist depth. Traditional Japanese
graphic art, one of the most advanced the world has
known, never departed from a resolute acknowledge-
ment of surface, despite the schools of Chinese painting
that continually thrived at its side.
It was not until the eighteenth and especially the
nineteenth centuries that the print-makers of Edo (not-
ably Hiroshige and Hokusai) introduced linear perspec-
tive, no doubt as part of the movement toward greater
'realism' linked with the rise of the merchant classes. I. In Ways of Thinking of Eastern
However, even a cursory examination of, for example, Peoples.
The Fifty-three Stages of the Tokaido by Hiroshige shows 2. Including, I would add, Christianity,
that the function of linear perspective within the frame- which clearly attracted the Japanese
because of its ritualized emotionalism
work of ukiyo-e was never predominantly illusionistic in and its erotic content. The Portuguese
the Western sense. It was primarily a procedure for the Jesuits were deeply perturbed by their
fifteenth-century converts' enthusiasm
articulation of the surface, only secondarily a rendering for the sado-masochistic aspects of
of the eye's perceptual production. This impression of Christian self-denial.

91
CROSS-CURRENTS reversal of emphasis appears to be historically confirmed
by the excitement which these works were soon to cause
in Europe among painters who were precisely trying to
move away from the illusionist tradition in painting. A
similar reversal can be detected in the Japanese adapta-
tion of certain codes or fragments of codes of Western
filmic 'illusionism'. And the Japanese attitude towards
the 'perspectival production' of the motion-picture cam-
era lens is one of the most interesting examples of the
transformational system at work within our field of inter-
est.
The Meiji overcoat, with its kimono-sleeves and
Western-style skirt, still to be purchased every winter in
Japanese department stores, may seem a frivolous exam-
ple, but constitutes a practical, comfortable and hand-
some combination of Japanese and Western styles 3 and
demonstrates the manner in which these transformations
have affected every aspect of Japanese life.
These three attitudes - acceptance, rejection, adapta-
tion - co-existed, then, during the 1920s and 1930s in
Japan - and in particular with regard to the representa-
tional system of Western narrative cinema, as they did
throughout the country's history, whenever she was open
to foreign exchange. As we examine these two decades,
we shall come to see how these attitudes functioned,
thus gaining a better understanding of contemporary
J. As did urban Meiji dress in general, developments in the Japanese cinema and of the remark-
combining Western trousers and short
'hapi' coat, bowler and fan, watch-fob
ably intense contradictions which characterized Japanese
and wooden clogs (geta ). society during those two crucial decades.

92
9. Lines and Spaces

By the end of the First World War, the predominance of

the American film in the world market was complete. It
was, of course, the war itself which had enabled Holly-
wood to triumph over France, her principal competitor.
The American studios had continued to work at full
capacity, while those of France and her European allies
were obliged to curtail their production drastically.
Japan's commercial and political ties with the Western
Powers grew stronger through her token participation in
the Allied war effort (occupation of the German conces-
sions in Shanghai), her presence at Versailles and her
action on the counter-revolutionary front in the Soviet
Union. Before the war had ended, she was seeing a far
greater number of American films than previously, since
Hollywood was able to fill the gaps left by the French,
Italians and English on the Asian market as well.
As I have already indicated, it was during the war, too,
that Hollywood, through the efforts of such men as
Barker, De Mille and the Ince brothers, had brought the
silent narrative codes to what may be regarded as their
peak of perfection. American products had commercial
success in Japan, and Intolerance, shown in 1919, seems
to have made a particularly strong impression. The suc-
cess of films which clearly illustrated Hollywood's aban-
doning of the primitive mode of representation seems to
have been the main drive behind a short-lived but inten-
sive emulation of American production methods and nar-
rative techniques. It was around this time that the
Shochiku Company, which for some years had dominated
live theatrical production - kabuki and shimpa - turned
to the cinema. Bringing home several Japanese techni-
cians from Sessue Hayakawa's Hollywood entourage, the
firm launched into the production of films which were
resolutely 'progressive' in spirit and designed to compete
with the American product. Actresses were hired to
replace the still universally employed oyama, location
settings were used as much as possible, 'spoken titles'

93
CROSS-CURRENTS were introduced for the first time and there were even
hopes of eliminating the benshi. Above all, of course, the
'Griffith codes' of editing were used in accordance with
what were felt to be the Western norms. 1 However, even
in this period of keen attentiveness to the Western
model, the Japanese reading of the Hollywood codes
brought about, as we shall see, transformations similar to
those which had characterized shimpa vis-a-vis its
(imaginary) Western model in the r8gos. Similarly, the
attention paid to the West by the new film-makers was
comparable and indeed related to the Westward-looking
attitude of the Meiji advocates of liberal reform.
The 1920s saw the rapid development of class struggle
on several fronts. Immediately after the war came the
country's first major outbreak of industrial strikes, com-
pounded by a recrudescence of peasant agitation, as
Japan was hit by the world-wide wave of inflation. At the
same time, big business was fighting to wrest power from
the post-feudal oligarchy and democratize Japan. Its
chief allies in this struggle, despite secondary contradic-
tions, were the new liberal middle class (particularly the
intelligentsia) and a fast-growing proletariat. The attempt
to 'Americanize' Japanese films was in fact a direct con-
sequence of this situation, just as the first attempts to
Westernize the theatre had been linked with the political
struggles of the r88os.
The earliest manifestations of shimpa were known as
sushi shibai (political theatre). Its ideology and even its
1. The movement was also leading personalities issued directly from the largest
characterized by some remarkably
picturesque excesses: scripts written in
bourgeois political party, liyu dantai (Liberal Party). Sig-
roman characters, a Japano-American nificantly, the first of the new plays celebrated the man
director. Henry Kotani, who insisted on
using English to direct
who drafted the first Japanese Constitution, whose
uncomprehending actors (see Anderson promulgation was a victory for the liberal bourgeoisie.
and Richie, p. 42). Clearly this was a
phase of wholesale adoption.
Shimpa plays proper generally contain some mild social
protest, and the shin-geki movement was in large part the
2. Although only a few of the first
new-style films, made around 1920,
expression of a Left that was just beginning to organize
were ideologically related to the itself around the turn of the century. 2
shm-geki movement, it seems that the
acting techniques of that theatre are
Of course, all of Japanese literature had had a social
preserved in the few surviving content, in the sense of dealing with class relationships.
examples.
And many if not most of the new style films of the early
3· Curiously enough, it died at about the I 920s advocate in fact a very traditional fraternity: the
same time as the last performances of
rensa-geki were being given in remote
feudal allegiances of the Edo stage were perpetuated in
provincial towns, and this double portrayals of class collaboration. Such films were no
demise is typical of that period of
cross-currents. I should perhaps add
more revolutionary than the social melodramas of Grif-
that an important factor in the rather fith. However, as we shall see, some more truly radical
abrupt cessation of this first 'Western
schoo]' was the destruction of the
films were produced later in the decade as the strength of
Shochiku studios by the terrible KantO the working-class movement continued to grow.
earthquake of 1923, together with most
of the studios of Tokyo, where this
In its earliest manifestation, then, the Americanized
brand of film-making was concentrated. film was short-lived. 3 Not unexpectedly, the Japanese

94
audience was unwilling to' accept alien modes of rep- LINES AND SPACES
resentation, which they had hitherto applauded only as
exotica, a category which the Japanese tend more than
other peoples to isolate (as, for example, the 'Western
room'). Explicit pressures were also brought to bear.
These were no doubt political, to the extent that the films
reflected the aspirations of the liberal middle classes.
However, they were primarily corporate pressures, since
important professional groups such as the oyama and the
benshi were directly threatened by all the Western inno-
vations.
We have seen that the objectively progressive forces in
Japan at this time identified their struggle with the intro-
duction of the Hollywood codes and the elimination of
the benshi, oyama and other elements inherited from the
traditional arts and culture. It may therefore appear sur-
prising that the partial failure of this movement should
be regarded in the framework of this study as a positive
factor. However, it is a mistake, as Western history has
taught us, to establish mechanistic equivalences between
historical and artistic progress. The development at that
time of a specifically Japanese cinema, whatever its fail-
ings, was an essential condition for the emergence of
what was to be, objectively, a materialist approach to film
art. This is a contradiction one repeatedly encounters at
every subsequent stage of development of Japanese films
until the 1960s.
Of course, the non-concordance of historical and artis-
tic progress is a universal phenomenon. However, Japan
must be viewed as an extreme instance. For the struggle
between dominant and ascending classes has, since the
opening of the country and until only very recently,
always corresponded to a conflict between the objec-
tively materialist artistic practices of the native culture
and idealist dominant practices of the capitalist West.
There is, consequently, on both the theoretical and prac-
tical levels, an irreducible contradiction between the
reading of their own culture's accomplishments by the
most enlightened and progressive Japanese (especially
the Marxists) and a Western reading which takes into
account one of the major theoretical acquisitions of con-
temporary Marxist thought, i.e. the non-equivalence of
art and history. The full implications of this issue, central
to the present essay, will be realized when we examine
the cinema of the 1930s.
The active hostility of the benshi was certainly a major
factor in the failure of the abortive movement to
'Americanize' the Japanese film. As already pointed out,
they were hostile to complicated narrative forms, and a

95
CROSS-CURRENTS profusion of flashbacks and cross-cutting were essential
ingredients of the new style. The benshi were, moreover,
the mainstay of a mode of representation which the
Japanese audience overwhelmingly preferred. 4 The ben-
shi reached the height of their popularity just about this
time, a fact which is at least as significant of the social
status of certain modes of representation in this period as
of the supposed universality of the star system.
The benshi as an institution was, however, by no
means immutable from its inception at the turn of the
century to its demise in the late 1930s. The earliest benshi
worked, of course, mostly with the first American and
European films, many of which were documents, authen-
tic or fabricated. 5 They simply held forth as steadily as
possible explaining the obvious and the obscure with the
same unfailing verve, often repeating themselves imper-
turbably. By the end of the Meiji era (1912), however,
the spread of fiction films and the development of a
native industry had led the benshi to concentrate almost
exclusively on rendering the supposed dialogue. Stylized
sound-effects were also used (as in the West during the
hey-day of the 'lecturer', only to be quickly abandoned
when the cinema moved away from the side-show and
into the theatre). In the years that followed, as the tech-
nique of dialogue was mastered, the benshi began
increasingly to interlard comments 'of their own', and to
describe, in particular, 'wordless' scenes. In short they
would read the images aloud. Thus by the end of the war,
there had developed in the cinema a procedure closely
related indeed to that of the doll theatre, whose texts
similarly mingle description with dialogue, often in rapid
4· Richie and Anderson (op. cit., p. 39)
succession. 6
point out that 'the Japanese audience As I have indicated, the peak of the benshi's popularity
was not yet trained to expect perfect (c. 1918-25) corresponded with the post-war period of
illusionism in films' and that,
consequently, a convincingly feminine inflation and social unrest. Interestingly enough, it was in
oyama wearing heavy workers' boots 1918 that a form of censorship came to affect the twenty-
was a 'fault' which did not disturb them,
even in the context of these 'naturalistic' year-old institution. 7 It seems to have been requested by
films (which also continued to employ the most respected and high-minded benshi themselves,
the same actor in more than one role).
who felt that the good name of their profession was being
5. I am indebted for this information besmirched by unscrupulous parvenus who interspersed
also to Yoshida Tieo.
their commentaries with smutty and, one suspects, topical
6. See Keene, Donald (trans.}, The remarks which purportedly disturbed members of the
Major Plays ofChikamatsu.
audience. In any case, the police instituted a licensing
7. The films themselves had of course examination; henceforth a certain level of 'culture' and,
been the object of censorship from the
very beginning. This was to be expected presumably, of social standing, was required of a would-be
in a country where censorship had benshi. One of the results, by no means fortuitous, was that
existed in some form or other for
centuries and where it still applied to the number of benshi practising at any one time was
every form of entertainment and limited. Since it was the established benshi who instigated
expression including, of course, the
press. this regulation of their profession, it would seem to be
linked with the practice of dividing the roles of Japanese LINES AND SPACES
films among several benshi. This practice did not exist in
the doll theatre, but it had far more to do with the benshi's
employment problems than with any concern for greater
realism, at least at this stage.
Although the benshi, and the modes of representation
which he helped to preserve, constituted the major hin-
drance to the implantation of the Hollywood codes, he
must not be simplistically viewed as an obstacle in the path
of progress. In Japan, the successful introduction of the
Hollywood codes, in association, for example, with a very
different development of the political and ideological
struggle - in other words, with an impossible rewriting of
history - would have been artistically far less productive
than in the West. The hegemony of bourgeois ideology
over the modes of filmic representation did at least make
possible the reiteration/translation of the great forms of
our literary and theatrical past and hence the creation of
such undeniable masterpieces as Mabuse der Spieler, Nana
or Foolish Wives. No hypothetical tabula rasa of culture
could ever have produced the equivalent in Japan (we have
only to look at the fate of painting and music there). And
while it was to take another decade for the specificity of the
Japanese cinema to assert itselffully, that development was
determined in large part by this resistance to the Holly-
wood codes offered by the benshi and his allies - such as
the oyama and chambera - and their audience.
The Hollywood codes can, as I have suggested, be
described with a fair degree of accuracy as narrative codes,
not because they function only on the narrative level as
such, but because they have been so intimately associated
with a mode ofrepresentation dominated by a certain type of
narrative attitude. Their most characteristic function has
been to give linear organization to the signifiers, to the
objects of audience perception within a given pro-filmic
space, in other words to line them up in those ordered
chains which began, after the 'Griffith revolution', to con-
stitute narrative in cinema. This narrative was consubstan-
tial with the literal sequence of images and titles. Its mode
of apprehension was preordained. The spectator 'went in'
at one end and 'came out' at the other, having followed a
path determined with maximum precision, as univocal as
the procedures of editing, the organization of angles, shot
sizes and camera movements, could make it. Each of these
infrastructural signifiers had one and only one correspond-
ing meaning. Each was concatenated in a one-to-one rela-
tionship with its immediate neighbours in the chain and,
through these, with the totality of the chain as it stretched
into the 'past' and 'future' of the narrative.

97
CROSS-CURRENTS The new-style producers and directors shortly before
and after 1920 failed, as we have said, to gain acceptance for
their ideas among the Japanese movie-going public, failed
to transform the Japanese cinema according to the canons
of Hollywood and its European emulators. This failure is
imputable to the fact that this need for a linearization, a
re-ordering of the 'primitive' image in terms of a pseudo-
linguistic montage, was meaningless to the Japanese. It was
antithetical to Japanese art, literature and language. 8
Narrative as such is not foreign to Japanese tradition; it
is, on the contrary, omnipresent, but its modes are radically
different from ours. As we have already seen, in kabuki and
the doll theatre the primary narrative dimension is isolated,
set apart from the rest of the theatrical substance, desig-
nated as one function among others. In the West on the
other hand, since the eighteenth century, our major narra-
tive arts - the novel, the theatre and more recently the
cinema - have tended towards a kind of narrative satura-
tion; every element is aimed at conveying, at expressing, a
narrative essence. Even in much of nineteenth-century
opera, the musical text was regarded as an exteriorization
of inner states (feelings, unstated conflicts, memories, etc.)
implicit in the libretto.
There exists a large and ancient body of Japanese solo
narrative arts, sung and/or spoken, to which both the doll
theatre's gidayu bushi and the silent cinema's benshi are
closely related. The complete list, which also includes the
already mentioned naniwa bus hi and rakugo (comic mono-
logues), is far too long and specialized to be set down here,
but one of the oldest and most seminal instances is the
heikyoku. Drawing upon a vast body of verse which
describes the struggle for power at the end of the twelfth
century between the Taira and Minamoto clans, these lyri-
cal narrations employ a set number of typed melodic
phrases. The heikyoku singer, who accompanies him- or
herself on a stringed instrument called the biwa, almost
always sings and plays successively .9 The singing itself is
strictly divided into two types. The narrative singing per se
calls for the use of a 'white voice', one which is 'transparent'
to the factual narrative; the lyrical passages which consti-
tute commentaries, involve the use of a codified set of
'colours', i.e. elaborate vocal techniques. One of these
8. I wish once again to pay tribute to the three modes of discourse (chant as against song or instru-
work of Iwamoto Kenji, who arrived
independently at conclusions essentially
ment alone) has been set aside for the narrative function,
identical to my own in this matter. They the other two have clearly non-narrative functions. Here is
have been set forth in his unpublished
lecture, The Benshi and Montage in the
a horizontal prefiguration of the type of disjunction which
Japanese Cinema. was to achieve such radical perfection in the doll theatre.
g. See Ueda Makoto, op. cit.,
Of all the pre-Edo arts, however, the hand-scrolls of the
pp. I 14-27. Fujiwara (Heian) and Kamakura periods offer the most
remarkable examples of an art in which the narrative LINES AND SPACES
dimension is thus circumscribed and isolated. Each scroll is
related to some pre-existing text, of oral or literary origin.
It was the Heian period which saw the introduction of this
genre from China and the birth of a pictorial system of
representation which uniquely characterizes Japanese art.
These early scrolls consisted of separately framed images
alternating with text relating the corresponding narrative.
The effect is something like a strip cartoon with written
panels in place of balloons. The best-known example of this
form is the Tale ofGenji (Genji Monogatari) scrolls, prob-
ably dating from the mid-twelfth century. In the Kamakura
period, these texts disappear and the scroll becomes one
long, continuous 'narrative surface', divided into episodes
by various representational or quasi-representational pro-
cedures. In neither form, however, can one say that the
narrative is 'contained' by the imagery. These scrolls were
intended to be 'read' in conjunction with the texts and, at
the time of their making, they were. On the other hand,
their rhythmic articulation of the painted surface is sup-
remely sophisticated, offering an autonomous plastic
statement. The scrolls in no way seek to 'clarify' the text, or
to give greater 'presence' to the action or 'depth' to the
characters, in the manner of the great Western illustrators.
In the Tale ofGenji series and in the other Heian scrolls, it
is impossible to distinguish between faces; they are drawn
according to a system of dots and dashes which precludes
personalization. The scrolls are both autonomous and
redundant with respect to a narrative text; they depend
upon their audience's supposed familiarity with the litera-
ture and oral tradition of the culture. It was the custom to
read aloud in small groups the texts of the Heian scrolls
while looking at the images, but this type of combined
performance does not seem to have survived the disappear-
ance of the inserted texts. I nevertheless maintain that the
relationship between the scroll and narrative material is
essentially analogous to that established between the
gidayu bushi and the stage-craft of the doll theatre. Both
provide, to quote Barthes, 'a total but divided show'- the
doll theatre within the space of the theatre, the scrolls
within a vaster socio-cultural space.
It is the heritage exemplified by these practices that
determined the rapid rejection of the linear narrative codes
of the West by the Japanese film-going public.

99
 10. The Fate of Alien Modes

The films made in the new pseudo-American style were

predominantly films with a 'social content'; resolutely
contemporary, moreover, in setting, they generally reveal
efforts to emphasize the aspects of Japan which were vis-
ually Western. They depicted milieux in which people
predominantly if not exclusively wore Western clothes
and lived in Western houses. 1 The actors' make-up, bear-
ing and gesticulation were often modelled on those of
American movie stars. And of course, as we have seen,
these films adopted wholeheartedly the codes of decoup-
age of Hollywood and Europe.
Each of the above-mentioned 'borrowings' involves
without a doubt a specific transformation deserving spe-
cial analysis. In examining these films, however, I shall
confine myself almost exclusively to a consideration of
the ways in which the codes of decoupage were displaced,
condensed, etc. This choice may appear arbitrary. But
our understanding of the manner in which iconographic
codes - which convey messages through photographic
reproduction, through the perceptual simulation of the
real - is, in my view, still embryonic. Moreover, this
study is primarily designed to trace the development of
Japanese cinema in terms of its specific modes of rep-
resentation. Those modes do not exist independently of
the iconographic codes. These, however, will concern us
only insofar as they are involved in specific transforma-
tions of the basic representational system of Western
cinema. This system may be analysed as functioning
along three fundamental axes: surface/depth; centering
/decentering; continuity/discontinuity. These issues had
been 'resolved' within Western cinema in a direction
already indicated in Chapter 5. Our task now is to see how
they were dealt with during a similar period in Japan.
1. How 'faithfully' these were depickd I
Among the handful of films which have survived from
have no way of knowing with any this brief period, the best known is Souls on the Road
certainty. The scant evidence I have
seems to point to the existence of such (Rojo no reikon, I 921) directed by Murata Minoru and
thoroughly Westernized milieux. Osanai Kaoru. Since intolerance, shown in Japan two

roo
~r~\ oi \· i_'<J
CROSS-CURRENTS During the period (about one third of the film) when
the 'souls' are on the road, the narrative editing is par-
ticularly complex, moving back and forth along parallel
and/or intersecting 'time-lines' of several of the main
characters. We see, often in rapid succession, the present
and then, in mental flashback, the past of the son and his
wife on the road and of his father and mother in the old
homestead. The son's flashbacks, and those of his wife
and parents, tend to centre around the same events (in
particular the circumstances in which the son quarrelled
with his father and left home), so that we see the same
scenes from different 'viewpoints'. Furthermore, during
this part of the film, the present-tense scenes, laid in the
two houses - those of the squire and the poor farmer -
compose in a sense the 'future' of the two travelling
groups. There are consequently four present times, four
pasts and two 'futures' in this film; they are strung out
over the 'time-lines' of seven adult individuals and one
couple, the ex-cons, who never separate, either mentally
or physically. We are thus given ten different stations
in the space/time of the diegesis. 2 If this film is in any
way typical it is understandable that the benshi were
befuddled by the new narrative approach and that
the new directors, partly hoping to pre-empt the ben-
shi's function, should have introduced a profusion of
sequence-prefacing explanatory titles as well as many
spoken titles.
This first section of the film, remarkably symmetrical
in its organization of the elements described above, caps
that regularity with a chance meeting, at a crossroad, of
the two travelling groups. In the next section, which
almost totally excludes flashbacks (these return at the
end of the film), the cross-cutting is accelerated and the
suggestion of simultaneity is consequently more empha-
tic: the prodigal son returns causing heavy emotional
turmoil in the old homestead while at the very same
moment (but see below) the two ex-cons bungle an
attempt to burgle the squire's house, are punished by the
squire but then, at his daughter's charitable behest, are
taken on as gardeners and handymen. In these sequences,
the cross-cutting, of an intensity comparable to that
2. I would consider the sequences which
of the final section of Intolerance, achieves a degree
open the film and in which the squire's of spatio-temporal intricacy far more complex than that
daughter flirts with the poor farmer's
younger son, as a further (eleventh) of any Griffith film. Let us consider the opening of this
station, for while it is indeed an section in some detail:
association of the two future/presents
(the two homes), it is also a sub-plot, a
pure present in its own right, though at - the prodigal son's mother sees her son and his wife in
the same time implying another future,
since it is ultimately clear that this affair the distance, although still off-screen, presumably still
will have a 'happy ending'. struggling across the fields
102
- the ex-cons are preparing to steal food from the THE FATE OF ALIEN MODES
squire's house
- the prodigal son is now with his wife and mother inside
the house; 3 they hear the father arriving outside (cross-
cut with shots of father)
- the ex-cons stealthily open a window (cross-cut with
interior scenes showing preparation of the food which
they covet) (striking symmetry with the preceding se-
quence)
- the squire drives up in a cart and starts to unload it, a
direct echo of the very similar arrival of the other father
- in the poor farmer's house, father and returning son
are locked in tense confrontation (this scene is fairly
long, much longer than the implied time-lapse 'on the
other side')
- the squire, shotgun at the ready, captures the ex-cons;
a close-up of the squire cuts to ...
- a close-up of his counterpart, deep in a violent quarrel
with his son, inter-cut with shots of the younger son, the
mother and the servants listening in the next room (this
sequence is very long and we tend to forget the other
story)
- the devil-may-care squire's daughter is prancing about
in her devil-may-care way; suddenly she overhears ...
- the ex-cons and her father, who is still holding them at
bay with his shot-gun (there is no real sense of time-lapse
at all here). They lunge at the older man, and the action
is suspended on a close-up of the squire preparing to
defend himself
- poor farmer/prodigal son confrontation continues
- returning to the other action, we find one ex-con
thrashing the other under the threat of the squire's gun;
when the old man feels the punishment is sufficient, he
orders the two vagrants to change roles
- poor farmer/prodigal son confrontation continues
- continuation of whipping scene: the injured and unwill-
ing castigator is hardly able to hold the switch (all these
punishment scenes are inter-cut with shots of the 3· As it went without saying that the
daughter watching from a nearby hiding-place) mother's welcome would be one of
- the prodigal son sits down to table with his father warm devotion, this meeting is played
down to the point of obliteration.
(scene inter-cut with extra-narrative shots of animals
feeding- again an effect of symmetry) 4- I wish to stress that this transcription
is scene-by-scene, not shot-by-shot,
- the ex-cons now plead with the squire, invoking the which would be non-pertinent here. Nor
bad treatment received at farms along their way (illus- have I taken into account the numerous
dialogue titles. All the characters, of
trated flashback) to justify their act of desperation 4 course, have Japanese names; the
stereotypes used here are an attempt to
convey the remarkable resemblance, on
From this point onward, the frequency of scene changes the diegetic level, to the products,
is somewhat abated: the girl intervenes, pleads with her grounded in nineteenth-century
melodrama, of Western cinema of the
father; the prodigal son and his father argue over the same period.

I03
CROSS-CURRENTS dinner table; the ex-cons are (symmetrically) fed by
squire and daughter. The squire then goes to an inn to
invite other 'poor souls' to the Christmas party, complete
with tree, which we had seen his daughter preparing in
the first section of the film; in the meantime, the prodigal
son's intractable father has sent him out into the snow to
sleep in a shabby barn.
This bare account should make clear the manner in
which this film differs from its Western models. Its truly
hypertrophied cross-cutting is developed beyond Grif-
fith's wildest dreams. For the bulk of Western audiences
and film-makers, cross-cutting was still what the young
Griffith understood it to be - a means of showing alter-
nate aspects of a diegetically homogeneous action.
Constantly wavering between the cross-diegetic cutting
of Intolerance and the homogeneous system of Birth of a
Nation, Souls on the Road, through a varied and quite
subtly graded use of rhymes and other correspondencies,
sets up a system whose sophistication remotely anticipates
that of Strike or October. Souls on the Road is, moreover,
an excellent illustration of the inadequacy of the classical
distinction between 'alternate montage' and 'parallel mon-
tage', since it partakes of both: at one level, the film is
composed of two 'separate' diegeses. Yet these nonetheless
communicate with each other through peripheral charac-
ters and two chance meetings on the road, one near the
beginning of the film, the other near the end. 5
Western audiences were not yet equipped to deal with
such complexities and would not be for at least two
decades, notwithstanding the various European avant-
gardes. Whether or not Japanese audiences of that time
were any better equipped is open to question The highly
intricate plot structures of so many chambera and their
5· Christian Metz's category of the
modern derivatives indicate, however, that on this level
'alternate syntagma' encompasses all too the topological habits of thinking we have previously
such figures. including presumably the
extra-diegetic, metaphorical parallels
alluded to had their role to play.
developed by Eisenstein (final sequence In Souls on the Road, we also have a markedly symmetri-
of Strike) and others. This is a far more
useful concept than the
cal patterning of both pro-filmic events and narrative
alternate/parallel opposition (Film articulations. This manifest formalization of what in Grif-
Lar ~uage, pp. 102-5).
fith was a purely signifying figure also 'anticipates' the
6. See my articles on Lang and Dreyer in structured alternative to transparency at its most coded
Roud, R. (ed.), A Dictionary of the
Cinema (forthcoming).
which was soon to be offered by Lang and Dreyer. The
organization of the cross-cutting and other elements in this
7· The exaggerated dilation/contraction
of the two narrative time-streams and
film, though determined culturally and ideologically on a
the very strange use, in scenes I have not totally different plane, is not unrelated to that of M or La
described, of overlaid images implying
the menta] presence of a distant
Passion de Jeanne d'Arc. 6 lt too contains objective radicali-
character (the 'real' character speaks to zations of the Western narrative codes, a geometrization of
this visualization of his or her thoughts
as to a 'ghost') are also examples of a
certain signifi.ers which is beyond the requirements of the
significant licence. codes as defined by the corpus which they underpin. 7

104
 The attitude manifest in this film does not seem to have THE FATE OF ALIEN MODES
been a matter of individual style, particularly if we consider
the probability that, although this film was related to the
anti-benshi movement, the spatia-temporal shifts sug-
gested above 8 would have been too great a risk to take at
that time in Japan without the presence of the benshi. For
no matter how 'topological' the Japanese mind, we must
not forget that the codes were brand new in the decisive
context of the native film. Yet the Japanese director of
silent films never had to fear that the public might not
understand the story. Even when he felt, as Murata and
Osanai did, that the benshi should go, he instinctively relied
on his presence. It is true that given the massive introduc-
tion of plot articulations of this kind (flashback, cross-cut),
he might have to cope with the obtuseness of the benshi.
However, while it was perhaps benshi resistance that
nipped this new style in the bud, Kinugasa was, after all, able
to find at least one benshi for his 'surrealistic' Page ofMadness
and Mizoguchi to construct a silent narrative as extravag-
antly intricate as 0-Sen of the Paper Cranes - indications
that this obtuseness was not unanimous. What is essential
here is that throughout the 1920s and well into the 1930s,
the problem of the elimination of spatia-temporal
ambiguities at every level of articulation, from eyeline
match to time lapse, though central to the very existence of
illusionist codes in the West, never seems to have bothered
the film-makers of Japan, almost undoubtedly because of 8. At times so intimate that the 'seams'
between two parallel actions are not
the benshi, whose task it was to spell out the diegesis. On always immediately apparent, even to
the strength of the picture alone, scene changes (spatial the present-day viewer. Of course,
iconographic elements may have
and/or spatia-temporal breaks) are often indistinguishable facilitated this distinction for the
from simple shot changes within the scene. The lack of any Japanese viewer of the period.
However, the rigorous adoption of the
clearly intended ambiguity (such as one detects on occasion codes of decoupage by film-makers such
in Souls on the Road) suggests that elementary information as Gosho during the period when sound
was gradually eliminating the benshi
of certain kinds was not considered necessary to be con- does indicate that these are indeed
veyed by the picture. 'This is still the same set' or 'now we universally valid if the objective goal is
transparent representation as the West
are elsewhere', were statements that explicitly or implicitly has defined it. At the same time,
formed part of the benshi's discourse. 9 according to Iwasaki Akira, Souls on the
Road was a commercial failure (in
Knowing that the new-style directors universally scorned Japan Film Yearbook, 1936). Is this to
the benshi as a basic impediment of the old style, we may be related to the general resistance of
Japanese audiences to the
assume, as I have already suggested, that the plethora of naturalization of the Western codes or
inter-titles, both descriptive and 'spoken', also represented to the film's 'over-use' of them?
Probably to both.
an attempt to set limits on the benshi's activity, to channel
his reading in accordance with the notions of unity and 9· Inagaki Hiroshi is once said to have
remarked that the Japanese directors
linearity which were inevitably a part of the process of took to sound without any difficulties at
Westernization. It appears, however, that the benshi coun- all since they already had it in the
benshi. As we shall see, it is possible to
tered this move by departing as far as possible from the text take this to mean that Japanese
on the screen even while it was being projected, and this directors already knew that sound and
image were two separate parameters as
was often very far. Thus, throughout the period of inter- are stage-craft and chant in the doll
titled Japanese films, three texts were produced during a theatre.

105
CROSS-CURRENTS performance: that of the image itself, that of the written
reading of the pictures provided by those who had filmed
and edited them, and of course by the benshi's own oral
reading, with its musical accompaniment.
Souls on the Road, then, was not a 'straightforward'
adoption of the Hollywood system. This was inconceivable
only fifty years after the Meiji Restoration, although it has,
particularly since the Second World War, been achieved in
Japan with far greater fidelity. It was not, on the other
hand, the simple illustration of one of the typical transfor-
mation processes as I have tried to define them. The film is
dominated by a contradiction between the intent to illus-
trate the foreign modes and their involuntary radicaliza-
tion. The Hollywood codes are used with an extravagance
that subverts to a degree their illusionist function; they are
visibly inscribed in the text of the film. The superabun-
dance of titles, presumably aimed, as in the West, at render-
ing the film as unequivocal as possible, at pre-empting the
benshi's discourse, 10 controlling and 'conscripting' it, actu-
ally resulted in an increased fragmentation of the signifying
tissue, the substance of expression, thereby strengthening
its ultimate similarity to the traditional theatre. 11
It appears, however, that in this early period, and even
among the films of the new style, such an intensive use of
spoken and descriptive titles was exceptional. While the
popular cinema (primarily chambera, but also shimpa-
inspired modern dramas) was still totally without them,
other 'advanced' films struck what might be regarded as a
compromise between these two positions. However, as is
always the case in these specifically Japanese transposi-
tions, the result of the 'compromise' was more radical, from
our point of view, than either of the extremes.
I have had access to only one example of this type of film,
10. It is hard to say whether the
but several Japanese historians assure me that it is known,
elimination of the American and through eye-witness accounts and surviving fragments, to
European lecturer was a strategy
consciously adopted by directors and
represent an aspect of the period. Winter Camellia(s)
producers. Whatever the case, it was (Kantsubaki, 1921) was directed by Inoue Masao, 'a shin-
quickly achieved, whereas in Japan a
fifteen-year anti-benshi campaign did
geki actor recently returned from America', according to
not succeed in removing him, and it was Anderson and Richie, for whom, in unfavourable contrast
not until the definitive implantation of
the technological infrastructure
to Souls on the Road, 'it smelled too strongly of the stage'
necessary for the generalization of and 'was quite artificial' .12 Setting aside the pejorative
sound made his presence redundant
and, in a time of economic depression,
overtones, these assertions are perfectly correct and cor-
onerous, that he finally did disappear. roborate the exemplary ambivalence of this film, par-
I 1. We shall soon see how Kinugasa,
ticipating as it does of both shimpa and shin-geki -and yet
also wishing to dispense with the benshi, which owed its commercial success to the presence on the
took the opposite course, completely
eliminating titles, and he too ended by
screen of one of Japan's first movie actresses. Not surpris-
radicalizing the new modes rather than ingly, this 'theatrical artificiality' went hand in hand with a
simply reproducing them.
greater regard for the benshi and his role, so that Winter
12. Anderson and Richie, op. cit., p. 44. Camellia - whose narrative need not concern us in detail

106
here - employs very few inter-titles, several dozen at THE FATE OF ALIEN MODES

most. 13
Titles in the film are of two sorts. There are laconic
commentaries, no doubt related to the type of remark
which the benshi would be making (for example: 'A strange
guest', preceding the first visit of the miller's shy daughter
to the house of the well-to-do family). 14 There are also
'spoken titles' which fulfil much the same purpose as in the
West. However, many of these are superfluous as titles,
since the words projected on the screen are redundant.
They are what is known in linguistics as phatic speech,
messages which, in cinema, are easily transmissible through
pantomime. Conversely, long scenes of dialogue, where
only the general sense is decipherable from the image, are
totally untitled; here, clearly, the benshi was on his own.
One of the most remarkable instances of this apparent
reversal is provided by the two final scenes, in which the
homicidal father is first prevented from committing suicide
by a police inspector, and then, as he trudges down the road
under arrest, is overtaken by his grief-stricken daughter. 13. -This film also deals with a situation
Neither the long farewell dialogue between father and based on class relationships, with the
rich playboy compromising the only
daughter, nor the slightly shorter one with the policeman at daughter of an ageing miller, who
the moment of the arrest, involves a single title; only the ultimately murders him. There is a faint
echo here of the exactly contemporary
daughter's cry of 'Father!' as she runs down the road after Foolish Wives which attests, rather than
him is honoured by a perfectly useless title, useless accord- to any less than hypothetical 'influence',
to these film-makers' attentiveness to
ing to the Western manner of titling. Various explanations the thematic arsenal of Western
have been offered for this phenomenon. Film stars are said naturalism. It is also interesting to note
that the 'realistic' tnurder (of a superior
to have demanded titles to highlight their roles (a habit also by his inferior) was, judging by the
prevalent in Hollywood). In this film, however, the few present state of an otherwise
impeccable negative, censored (in a way
titles seem to be distributed quite evenly among the cast, which would have made the sequence
although the woman, Mizutani Yaeko, is (and apparently quite bewildering without the benshi's
helpful presence). This of course
was) regarded as the top star. It is also said that the benshi confirms that whatever their ultimate
had considerable say in the use of titles; they may have historical role as regards the
development of film-art in Japan,
thought titles deserved only trivial expressions beneath a benshi, oyama, etc. were objective allies
benshi's dignity. There is probably some truth in both these of the political establishment in so far as
they were hindering the rise of a cinema
'explanations'Y In any case, the benshi almost certainly which it sought to repress.
tended to ignore these titles, either by silence while they
14. Of course, such titles had also
were on the screen or by improvising a different, perhaps become common in the West, where
even contradictory, text. they may be regarded as the surviving
traces of the lecturer.
These titles, then, were completely redundant; they insti-
tute, in fact, a new substance of expression, writing in and 15. Richie and Anderson's only
comment on titling at this period is a
for itself, as graphic element, as opaque signifier, as a reference to distribution contracts by
gesture of pure signification, which takes its place between which producers were 'reimbursed on a
direct per-foot basis.' This encouraged
the discourse of the benshi, pre-empting the other titles of them to 'introduce long and
their narrative function, and the image, pre-empting these meaningless titles, so much cheaper to
shoot than live action' (op. cit., p. 61).
titles of their phatic function. This mutual pre-empting of This may explain in part the profusion of
narrative functions could only heighten the divorce be- written text in a film like Souls on the
Road. It certainly doesn't account for
tween screen and voice. Here we have a configuration the rare and 'useless' titles in Winter
doubly homologous with that of the doll theatre. For, Camellia.

I07
CROSS-CURRENTS extending Barthes' analysis, one may equate the effective
gesture of the puppeteers' hands with that of the callig-
rapher's.
Winter Camellia complies in every other important
respect with the codes of representation that were by then
totally dominant in the West. 16 In particular, and in con-
trast with the then still prevalent character of most Japan-
ese films, considerable pains were taken to film location
and even some interior settings at oblique angles, a pro-
cedure17 which has been a strategy basic to the system guaran-
teeing the visual illusion of depth (as opposed to its mental
illusion, which hinges rather on the reverse field). The
oblique angle had been the rule in the West ever since the
film-makers of Griffith's day had succeeded in proscribing
the frontal viewpoint, immediately inherited from the
theatrical side of cinema's ancestry.
It should be emphasized that in chambera, the dominant
genre of the period, oblique angles remained exceptional,
despite the fact that when visual depth-effects were intro-
duced, it was in connection with the sword-fighting scenes
precisely typical of such films. Makino's Jitsuroku
Chushingura (The True Story of the Loyal Retainers, 1928)
is certainly no chambera, but a long series of hieratic
dialogues, filmed for the most part in a two-dimensional
spirit (see Fig. 3). It ends nonetheless at the moment of
Kira's punishment, with a bravura sequence in which fore-
ground and background actions are skilfully opposed. The
fact that the film was clearly a more elaborate, costly pro-
duction than the period average may have stimulated this
development. In any case, that last sequence is possibly one of
the earliest in which the aptitude of Japanese architecture
to produce a certain type of depth-effect is systematically
exploited. More generally this dichotomy was charac-
teristic of the average chambera, in which dialogue scenes
were uniformly shot frontally 18 while the fights usually took
place on location and by the mid-1920s were frequently
I 6. An interesting, though indeed filmed from all angles.
incidental exception is the use of
protracted long shots of a revolving
Between 1920 and 1926-7 the wide-angle, frontal shot
water-wheel as a mode of punctuation remained the rule for both interiors and exteriors, though
closer to the 'pillow shots' of Ozu (see
Chapter I6) than to the cutaway
the concertina and reverse field did begin, very slowly, to
'bridges' of Hollywood and Europe. spread, so that medium close-ups and close-ups became
J7. One might well call it a rule: whether
more frequent. However, it was not until Makino's matur-
trained in a school or a studio, our ity as represented by this late Chushingura that the West-
budding director or cameraman always
learns never to set the camera up at right
ern codes made real inroads on the serious period drama
angles to a flat background. which was growing out of chambera largely due to the
I 8. This flatness derived quite directly
impetus of Makino. Previously t0 that, I feel that on the
from the conception of the kabuki whole the use to which chambera put the concertina is,
'interior': an open-front box of scarcely
any perceptual depth set up here or
again, closer to the 'stage-economy' of kabuki than to the
there inside the proscenium frame. typically Western project of characterization/singulariza-

108
 11. Displacements and
Condensations

The best-known director of chambera during this period

was Ito Daisuke who was still making films until well
after the Pacific War (as the Japanese call their part in
World War Two). Though it would be wrong to say he
was the inventor of this genre, he was certainly one of its
major codifiers. Chambera grew out of a classical
encounter between Japanese presentationalism and a
demand for realism, stimulated by the impact of Western
ideologies and modes of representation, but produced by
the objective development of Japanese society. The chief
difference between the earliest chambera and the trans-
posed kabuki and shimpa plays of the previous two
decades was the greater realism of the sword-fights.
When a character died, he no longer performed a back-
ward somersault but collapsed in bloody agony; above
all, the actor's body struck the tense poses of actual com-
bat as opposed to the stylized, relaxed gestures of the
kabuki encounter. The word chambera, onomatopoeia
representing the clash of swords, very precisely desig-
nates this difference, since on the kabuki stage swords
never touch. 1
Aside from this important change and the gradually
increasing but still perfunctory use of the narrative edit-
ing of the West, the bulk of these films, as I have said,
remained very similar to those of the previous period,
with a predominance of head-on long shots and with the
narrative burden still devolving on the benshi.
Ito seems to have made several interesting innovations,
whether or not he did so thinking to emulate Western
procedures directly. He seems to have been the first to
1. At the same time, it seems significant introduce 'dynamic' editing into his combat scenes. He
of the place assigned to realism by
Japanese culture, that a verbal
initiated, reportedly at the behest of his skilful camera-
stylization of a sound should give its man, Karasawa Hiromitsu, that extravagant use of the
name to a silent film genre, in which the
sound itself was never heard but was
hand-held camera which was to characterize a whole
replaced , , , precisely by the school of chambera, from the mid-rg2os until at least the
onomatopoeia of the benshi and by
written (sometimes animated)
implantation of sound. An examination of these traits of
onomatopoeia in the titles, the 'Ito style' will provide us with some insights into the

IIO
problem of the transformation of the Western codes dur- DISPLACEMENTS AND
ing this period of Japanese assimilation. CONDENSATIONS
At present, the only known surviving samples of Ito
Daisuke's own art are a large number of fragments col-
lected by Matsuda Shunji. 2 However, a film preserved in
the Film Centre of the Tokyo Museum of Modern Art,
The Red Bat (Beni Komori, I93I) by one Tanaka
Tsuruhiko, clearly belongs to the Ito tradition. Since it is
one of the few intact examples of the mature period of
the silent chambera, comparisons are difficult. It is cer-
tainly related to the Ito fragments I have seen, and I have
been assured by Sato Tadao 3 that it is typical of this
important variety of chambera. 4
This film retains few of the pertinent traits of the
Japanese Primitive and what I would call the Neo-
Primitive periods, apart from that typical Japanese pre-
dilection for the long shot and the frontal view. However,
within a system of representation close to the Western
model, it introduces a number of 'displacements and
condensations' 5 which may be said to undermine deci-
sively the basic purpose of those codes not only as articu-
lated in the West but as faithfully adopted by those 2. The survival of so many fragments
Japanese directors who were impressed by their commer- from that period and so few complete
cial success, and/or experienced an artistic and ideologi- films is something of a mystery to me.
Presumably it is related to the practice
cal need for them. of compiling excerpts to make up for the
scarcity of films just after the war. This
I will not attempt to set forth the rambling plot of The practice was facilitated of course by the
Red Bat, which relates the swashbuckling adventures of visual similarity of so many films from
that period.
an impetuous do-gooder named Chotaro who tries to
repair the injustice done a jilted maiden. Our concern is 3. He has written, in collaboration with
Yoshida Tieo, a history of the genre,
with the strategies which Tanaka, closely following Ito's Chambera eigashi.
lead, borrows from the Western representational system,
wherein they already played extremely circumscribed 4· There is some internal evidence that
the film is, in part, meant as a parody of
and/or exceptional roles; our concern is with the way he the Ito style, and this aspect of the
project may be responsible for some of
converts them into polyvalent signifiers, as general as the the paroxysms described. Responsible
cut itself, insofar as it can acquire any of the different in degree, perhaps, but not, I am
convinced on the evidence of surviving
possible spatia-temporal significations according to the Ito fragments, in kind.
diegetic context. 6 These three strategies, in order of
importance, are, the swish-pan, the lap-dissolve and 5· At present I am unable to provide
theoretical justification for the use of a
'dynamic editing', which I will pragmatically and provi- metaphor borrowed from Freud's
analysis of the dream-work. I do feel,
sionally define as a succession of unusually brief shots. nevertheless, in dealing with the
The swish-pan (and more generally the free-wheeling relations between such radically other
cultures that the Interpretation of
camera movements characteristic of the Ito school) pre- Dreams provides keys to some of the
sumably came to Japanese films when the first small strange visions they will have of each
other.
cameras of American and French manufacture reached
Japan in the mid-I920s. Now in the West, until the 6. Spatial and/or temporal continuity,
spatial contiguity, temporal
spread of hand-held techniques in television news- simultaneity, spatial alterity, definite or
coverage and cinema-verite, the swish-pan was an excep- indefinite time-lapse, ultimately
tion and was always used connotatively, with what may flashback or temporal overlap. See
Some Terminological Indications; also
be called an exclamatory function, while its denotative my Theory of Film Practice, Chapter 1.

III
CROSS-CURRENTS value was almost always that of temporal continuity and
spatial contiguity. 7 This sporadic, specialized role is still
attached to the swish-pan in most narrative film.
In The Red Bat, the swish-pan occurs on the contrary
with extraordinary frequency, and is used to fill, alterna-
tively but not invariably, three different signifying func-
tions which in the Western codes had long since devolved
upon the cut, black-out or dissolve: reverse-field match,
cross-cut (cutaway) and flashback. As I say, this practice
is not invariable, for while many reverse-field series are
articulated by swish-pans, these often alternate within the
same series or from one series to another, with lap-
dissolves - another anomaly with respect to what was
by then standard Western practice - or plain cuts. Sim-
ilarly, while most parallel action sequences are linked
through swish-pans, simple scene-changes, often involving
a mode of spatia-temporal shift semantically equivalent
to the cross-cut, are only occasionally subject to this
7. Frequently today. exceptionally procedure. 8
during the 'classical' period of Western
cinema~the swish-pan has been assigned
The Red Bat, like, apparently, many films of the same
an unexpected denotative value school, extended the denotative polyvalence characteris-
(flashback, time-lapse and/or spatial
alterity). Of course, the
tic of the cut alone within the Western codes to other
connotative/denotative opposition is an types of shot articulations which, in the West, were
over-simplification, since we are
actually dealing here with a set of
pruned, as it were, of their polysemous potential and
specifiers which have meaning only assigned a single, undisputed denotation/connotation
insofar as they refer to the complex of
codes which constitute the diegesis: a
set. 9 The connotative functions are effaced entirely; the
cut, a swish-pan have no semantic value swish-pan loses its exclamatory value and the lap-dissolve
whatsoever in themselves, as does a
word which, even out of syntagmatic
its poetic implication. And while the intended meaning of
context, has a lexical meaning that has any given dissolve, swish-pan or cut in The Red Bat is
no equivalent here. However, on the
level of this 'meta-code', it does seem
ultimately specified by the diegetic context, as in the
legitimate to employ this terminology, Western film, the process of specification is no longer the
since the difference between the two
functions at stake can be shown to be same.
homologous with that which separates This free polyvalence of the three articulative signifiers
linguistic connotation from denotation.
most frequently used in this film is directly at odds with
8. In such instances, the common what was already the irreversible tendency of the dominant
practice is a variant on the
'cross-cutting' procedure of lateral
Western cinema: an increasing specialization of all 'special
swish-pans: the camera streaks upward effects' and an increasing polyvalence of the plain cut
to the sky; this shot is followed by a
similar pan down from the sky to an
alone, in conjunction with a sophisticated refinement of the
action distant in diegetic time or space. specifying powers of the diegetic organization. In contrast
9· It was during the period of conversion
with the historical tendency towards economy in the West-
to sound in the West that the dissolve ern codes, there is a kind of non-signifying residue in the
became equated with time-lapse and/or
spatial alterity. This development
swish-pans and dissolves of The Red Bat. A cut and a
responded to a need for specification swish-pan may have the same semantic value; moreover,
which had until then been satisfied by
the newly proscribed inter-title, in their choice as signifiers of, say, spatial alterity, is clearly
conjunction or not with the iris or arbitrary with respect to the narrative and precludes any
black-out. The Japanese extension of
the lap-dissolve was, in Western terms, a
'expressive' codification. Thus, these particular forms of
'restoration', since this procedure had shot-change carry a surplus mark which is extraneous to the
originally been far more polyvalent than
it was to become in the late r 92os and
process of producing meaning at the primary, spatia-
early 1930s. temporal level of the narrative.

II2
 Dominant ideology, in its relevant manifestations, qual- DISPLACEMENTS AND
ifies such non-signifying differences either as signs of CONDENSATIONS
'stylistic coherence' or as 'formalism', according to the
social prestige attached to their contexts. These value
judgements have no pertinence here. The Red Bat would be
dismissed as 'trash' by most critics, with the possible excep-
tion of those who habitually make a fetish of popular cul-
ture. These films are indeed part of popular culture, but in a
sense inapplicable perhaps in the West, since they derived
their partly presentational approach from conceptions
shared by an entire society for centuries. In the West, it was
a dominant class which, in the course of its struggle to
subdue a rising class, imposed upon that class (and ulti-
mately upon most of society) a synoptic version of what
were uniquely its own modes of representation. Most
important of all, these were watered down to an
impoverished functionalism in support of archetypal fan-
tasies and an ideologically weighted illusionism. The silent
films of Cecil B. De Mille are worlds removed from the
novels of Dickens. The Red Bat was still very close to the
Edo stage.
There are many types of articulations that can be used to
join shots in a film. It is, of course, no accident that the
briefest and the least perceptible assumed the basic func-
tion in the Western system of a highly developed diegetic
continuum. Aside from its greater technological simplicity,
the cut proved from the outset to be the ideally transparent
signifier; its invisibility- through editing procedures such
as eyeline matching or cutting on movement- was such that
the 'meaning', the message, seemed to issue forth naturally
from the context as a whole, from the flow of images itself.
In other words, the signification of this articulation receded
to a pre-conscious level. On the other hand, such articula-
tions as the iris, one of the earliest to be coded, no doubt
because of an evident analogy with the spotlight iris of the
stage, or the dissolve, uncoded until the 1930s, were both
experienced as disruptive, and continue to be so today.
They were perceived as signs, whereas the cut was very
quickly seen as 'change' or 'shift' or 'passage of time' as
such. This very material event came to be the transparent,
'immaterial' representation of certain narrative dimen-
sions, a representation as 'natural' as language. It was only
after much trial and error, that an adequate code of 'punc-
tuation' succeeded in locking the other shot-change proce-
dures into a comparably 'natural' system. This system
remained very fragile, always threatening to break down,
precisely because of the duration and the positive, per-
ceptual materiality of the signifiers involved: iris, fade,
dissolve, wipe. This frailty explains in part the frequent

113
CROSS-CURRENTS redistribution of functions over the years among these signi-
fiers, as one device becomes 'old-fashioned' (i.e. it is per-
ceived as material signifier) and is superseded by another.
Thus we may explain the gradual disappearance of the
wipe during and after the Second World War, the more
recent tendency to reserve the dissolve for special, expressive
effects and to convey ellipsis through a combination of the
plain cut with diegetic strategies. There are, however, no
grounds for what is commonly regarded as the 'natural'
distinction between the cut and, say, the dissolve. A space
between words is a sign as much as a hyphen or a period.
In contrast to the cut, procedures such as swish-pans and
dissolves, particularly when first introduced and still
uncoded, were visible as such precisely because they
impinged upon the representational integrity of the photo-
graphic image - of the iconographic codes. They were an
impediment to the conviction that what we see on a motion
picture screen is 'natural', a recreation of the conditions of
actual viewing. For this reason, by the time the introduction
of synch sound had completed the equipment of dominant
cinema, it became urgent to assign these procedures, nar-
rowly limited denotative and/or connotative functions, to
create automatic reflexes which would efface them. They
had, in other words, to conform to the linear model which
ordered both the reading of the iconographic elements
within the frame and the succession of the montage-pieces
as such. It was essential that there be no superfluous,
'gratuitous' marks which could call into question the
economy of this system: the bourgeois codes are thrifty. 10
As we have seen. this approach to representation was
completely foreign to the Japanese performing, literary
and visual arts. Yet at the same time, the editing system of
The Red Bat (and of the Ito scraps available to us) contains
the Western system, which very often functions quite nor-
mally within it. The approach here is, in fact, non-
committal. It refuses to constitute a system of its own. It is
the ever-present possibility of emergence/submergence of
the Western system which makes these films appear so
'deconstructive' of the codes that dominate our cinema.
The articulative strategies are often presented as such, but
very often they are not. The discourse shifts unpredictably,
arbitrarily from a 'hieroglyphic' to a 'phonetic' system -
which, of course, is one of the faculties of the Japanese
language.
One can easily imagine a film made entirely according to
Western norms in which, however, the cut would be sys-
tematically replaced by the swish-pan, and the dissolve by
10. For further remarks on this concept
two feet of red leader. We would still be dealing with a
of bourgeois thrift, see Chapter 19. perfectly linear code of discontinuities and once the code

114
had been learned, it would become perfectly 'transparent' DISPLACEMENTS AND
for users. It is the fact that in The Red Bat a given set of CONDENSATIONS
meanings are associated with three interchangeable and
'arbitrarily' chosen signifiers which allows us to claim that
this film does not simply substitute one code for another
but constitutes, on this narrow but significant level, a chal-
lenge to the very notion of the code. It is true enough that
this same challenge was set forth far more coherently in
Europe the very next year (1932) by Dreyer's Vampyr.
Dreyer, however, as we know, had already entered the
forty-year purgatory when he would be allowed to make an
average of one film every five years, seen only by an elite
audience. Directors like Tanaka and Ito annually made
dozens that were seen by the entire Japanese film-going
public. What was a mass cultural attitude in Japan was a
deeply subversive vanguard practice in the Occident.
The third Western strategy to be adopted and trans-
formed by this school of chambera was that which I have
schematically termed 'dynamic editing'. It is characteristic
of the silent period of Soviet cinema and of Eisenstein's
films in particular. It also characterizes some early films of
the first French avant-garde or 'Impressionist School'
(Gance, L'Herbier, Epstein). Together with the 'Paris Rus-
sians' (Volkoff, Mosjoukine and Kirsanoff) - who took
most of their ideas from the experiments of Soviet and
French pioneers - it seems it was the latter who were
directly instrumental in acquainting Japanese directors
with the European movement's reaction against the impos-
ition of the system of representation that had been per-
fected in the United States.
The decoupage of certain sections of Potemkin (I, u, IV)
illustrates a critique of dominant cinema on the part of
advanced film-makers in Eastern and Westerri Europe at
that time. Simultaneously, other film-makers in the Soviet
Union and elsewhere, from Pudovkin to Volkoff,
wholeheartedly committed to the ethos of representation,
illusion, expression, identification, etc., were employing
'fast cutting' to express 'violence', 'passion', 'music',
'dance', 11 or otherwise endowing it with meaning, reducing
it, in fact, to a monovalent signifier. This is the form in
which 'fast cutting' was eventually assimilated into the
codes of dominant cinema throughout the West: a special-
ized, connotative element, much like the swish-pan. 1 I. Of course, such non-verbal signifiers

This coded approach to 'dynamic editing' soon found its cannot be reduced to verbal ones. The
signifying process in film is homologous
way into the Japanese film, where it often appeared side by to, but not identical with, the linguistic
side, however, with the polyvalent approach to other pro- process proper. I might also add that
Eisenstein strayed into the paths of this
cedures as already described. In Makino's 1928 Chushin- 'expressivism' as well, not merely in the
gura, swish-pan and dissolve often have that nominally writings of the 1930s, but in the films of
the 1920s (the final section of Potemkin,
'neutral' function, equivalent to that of the coded cut, the Tartars' dance in October).

115
CROSS-CURRENTS tending only intermittently towards a 'Western' connota-
tive function (exclamatory, 'poetic'). At the same time, all
the moments of greatest tension during the various meet-
ings of the ronin (to discuss the ethical problems involved in
the pre-vendetta situation) are systematically under-
scored by an acceleration in the succession of medium
close-ups which predominate in these scenes.
The chambera of the Ito approach also follows, to some
extent, the dominant Western model. Scenes representing
physical violence are similarly intensified by fast cutting
and free-wheeling (hand-held) camera movements. A
sequence near the beginning of The Red Bat in which
Chotaro demonstrates his prodigious strength by hurling
cart-wheels and various farming implements high into the
air and on to his numerous assailants, is an excellent
instance of a type of scene which seems to have abounded
in Ito's own films as well as those which he influenced.
However, this scene also reveals a contradiction which
typifies the Japanese attitude at that period. For as it
unfolds, it also results in a playful but nonetheless radical
decentering of the imagery with respect to the narrative
situation, a weakening rather than a strengthening of the
diegetic effect. The introduction of vertical, lyrical
developments of plastic and kinetic motifs brings the diege-
tic flow to a standstill (as in Eisenstein's The General Line,
for example). At one point during Chotaro's exhibition-
which in itself has the 'presentational' quality of an aragoto
('rough stuff') dance in kabuki - we abandon the earth-
bound protagonists and onlookers completely for nearly a
minute, to observe a pyrotechnical display of flying cart-
wheels, harrows, etc., seen as black, abstracted silhouettes
against a white sky. The effect is one of vertical accumula-
tion of empty signs, a momentary negation of the diegetic
plenitude from which they are nonetheless excerpted. And
one may say that, starting from an attitude implying
wholesale adoption of a Western code, a typical Japanese
process of transformation and radicalization (traceable,
through benshi, kabuki, etc. to the historical modules)
issues, spontaneously, as it were, in the critical modes
through which advanced Western film-makers challenged
the domination of bourgeois illusionism.

II6
12. Surface and Depth

We know that the modes of representation of volume

and depth as developed in the visual arts between the
thirteenth and fifteenth centuries constituted a fresh
point of departure for figurative technique. This
development was also an important moment in the intel-
lectual history of the West. One of the main obligations
incumbent upon those who later undertook to codify the
modes of filmic representation in accordance with the
values of the dominant class was to implement a filmic
response to the need for the rendering of depth. 1 Con-
versely, the general lack of concern with visual depth in
the arts of the East and particularly those of Japan (e-
makimono, or picture scrolls, Muromachi screens, the
prints of ukiyo-e; no, kabuki, doll theatre) were preserved
in cinema long after the 'laws' 2 of depth-representation by
the camera were established in the West.
The acknowledgement of the pictorial surface in
Japanese painting and the tendency of Japanese poetry
to focus the reader's awareness on the graphic and textual
surface may be seen as perfectly antithetical to the two-
fold matrix, visual and dramaturgical, which was imposed
I. It is not true that the machine itself,
on the cinema in the West during the Formative period. the camera, 'produced' from its very
As we have already noted, some of the major efforts to inception, a linear perspective in happy
conformity with the needs of that
set the Hollywood system at a critical distance (Godard, bourgeoisie whose science had
Dreyer's Gertrud, Warhol, Snow) have been directed produced the machine. The notion of a
bourgeois science is no more valid in our
against its modes of depth representation. The very first field than in genetics. We need only
film to undertake systematically such a task, The Cabinet compare what was so often the patent
flatness of 'Primitive space' with the
of Doctor Caligari, returns to the flat frontality of the developments which gave the
Primitive cinema in conjunction with a very sophisticated depth-producing strategies the status of
a system.
exploration of the depth/surface and 'opacity' /'trans-
parency' dichotomies. In Caligari, strategies of relief 2. I have already referred to the
hard-and-fast 'film-school" ban on
(oblique trajectories, perspective convergency) are intro- filming at right angles to a wall. One
duced within a pictorial system whose painterly flatness might add the equally stringent rule
against 'zooming' in such a position.
contradicts and in fact explodes the conventions which had These laws also include the complex
by 1919 developed to an unprecedented degree of effi- codes manipulated by the lighting
cameraman and set-designer, directly
ciency in the West. For setting aside the complex, though derived from the principles of
ultimately trivial, problems of 3D, it may be said that by post-Renaissance perspective.

II7
CROSS-CURRENTS the time the dominant mode had been completed on the
visual level, thanks to the establishment of the 'envelop-
ing reverse field', the cinema had succeeded in acquiring
the illusionist strategies of painting, sculpture, theatre
and literature as perfected during the nineteenth century.
All that was missing was synchronous sound.
The flattening of the image in the work of, say, Ozu
after 1933 may legitimately be regarded as a deliberate
'throwback', comparable to the strategy of Caligari. The
persistence of primitive flatness in so many Japanese films
during the 1920s is, on the other hand, 'accidental': it was
a collective phenomenon and came about through the
interaction of certain Western editing methods with cer-
tain characteristics of traditional Japanese architecture.
Yet, insofar as it was determined by the general 'surface
orientation' of the arts of Japan, especially of her
architecture, 3 it was, of course, no accident at all. Nor
was Ozu's strategy a purely personal invention.
When the traditional-minded Japanese director of the
1920s began moving his camera in for increasingly fre-
quent medium and close shots, he felt no need what-
soever to modify the 90° angle of camera to background.
(I am referring, of course, primarily to interior sequences
or to exteriors involving architectural planes.) Nor did
this sparing introduction of editing into most films dimin-
ish in any way the dominion of the frontal view and the
long-shot (in marked contrast, by the mid-1920s, with the
decoupage of the average Western film). And, of course,
these wide-angle shots allowed the native architecture to
assert itself fully. The significant absence of furniture in
the Japanese interior, the comprehensive articulation of
every surface in terms of asymmetrically disposed
rectangles (the purest expression of the concept of two-
dimensionality) inevitably went a long way towards pre-
cluding from the master shot the issue of illusionistic
depth. This being the case, the Japanese presumably felt
that the introduction of depth-producing oblique angles
in the close-ups, even though this was the rule in the
Western context, would destroy what was for them an
important factor of unity.
This resistance to a key element of the Western system
suggests something other than a passive traditionalism; it
corresponds to an essential, active component of the
Japanese attitude towards representation. Compare it,
for example, with those uniquely Japanese picture gar-
dens, laid out in such a way as to be visible only from one
vantage point and to create the illusion of pictorial flatness
in a three-dimensional field.
3· See Chapter r8. A wall of plaster or board generally enclosed three sides of the

II8
plot: within this frame, the gardener 'painted' a picture, using SURFACE AND DEPTH
stones for brushstrokes ... To compensate for the distortion
brought about by having the 'picture plane' stretch away from
the observer along the ground, predominantly vertical shapes
were grouped at the rear and the ground itself was often
pitched downward imperceptibly toward the veranda, from
which the garden is seen. At Daitokuji temple, in the garden of
Diasen-in ... there is a white plaster wall, now half-hidden, that
once carried the tone of the sand-strewn ground up behind all
the clipped shrubs, thus flattening on to an imaginary picture
plane objects in reality some distance from each other. 4
At the same time, as a supreme sophistication, the
graphic techniques of (Chinese) perspective were re-
introduced, as it were, into the 'picture plane' to indicate
a scale many times that of the actual garden.

At Diasen-in too ... because the waterfall is supposed to be at

a great distance away, the trees behind it are smoothly clipped:
the stone bridge which indicates the middle distance is much
larger in scale, and trees in the foreground, also larger in scale,
are left untrimmed so they may have the detail one expects
near at hand. 5

This dialectical commingling of Japanese and Chinese

modes of representation is an example of what so often
results from such 'borrowings', a further refinement and
radicalization of the dominant element in this union: the
Japanese system which we have already seen at work in
so many areas, that 'system of surfaces' which pervades
all of Japanese thought and life.
As previously mentioned, aspects of Japanese architec-
ture that might have contributed to a 'Westernization' of
the Japanese approach to pro-filmic space, seem to have
been almost systematically neglected by most film-makers
until the very late 1920s. One notes, in particular, the
neglect of those successions of communicating rooms,
which in the future were to play such an important role in
the development of sophisticatedly ambiguous depth-
surface relationships. 6 I have already cited a noteworthy
exception to this rule, Makino's 1928 Chushingura, which
contains, besides the spectacular finale, numerous, briefer
instances of oblique 'deep-field' shots, especially in the
opening sequence, evocative of Mizoguchi's future master-
piece. Apart from this film, the very exceptional work of 4. Drexler, Arthur, The Architecture of
Kinugasa and the films of Futagawa Montabe (the epitome Japan, p. 176.
of the academic neo-Western director), the depth potential 5· ibid .. p. 179·
of Japanese architecture is neglected in all of the other films
6. Budgetary considerations no doubt
of the 1920s that I have seen. Even the faculty of the shoji often played a role in this
(sliding partitions) for revealing sudden depth as they are over-determined phenomenon.

119
CROSS-CURRENTS drawn back, widely explored in later years, was enlisted in
aid of the general emphasis on surface. When a shoji fills
the frame at the beginning of a shot (first perceived less as
'door' than as detail of the general rectangular pattern),
then opens to reveal a character about to enter a room, the
depth which we know to have been revealed by such an
action is contradicted by the head-on, rectangular flatness
of the final image, with its neutral, 'distance-less' back-
ground behind the human figure framed in the opening. An
expectation of depth is met by flatness. This effect is
directly related to a procedure employed in kabuki on-
stage interior sets: a sliding door is drawn back by an
unseen hand to reveal a new tableau, adding a rectangular
vignette to the picture plane of the stage. And often in these
early films the framing is such that there is the same feeling
of separation between the actors' appearance and the open-
ing of the shoji; it is almost as if a stage-assistant had
performed the effective gesture. In the films of the 1920s,
this type of entry soon became fairly frequent, together
with the complementary mode of exit. The shot either
began or ended with an 'empty frame', as the camera
focused on the austere rectangles of the exposed wooden
ribbing. The 'wipe' effect by which characters thus
appeared or disappeared was a further factor in confining
these ambiguous images to the two-dimensional screen
surface. The systematic use, by the mature Kurosawa, of
the hard-edge wipe is a significant echo of this practice.
The acknowledgement of the surface is most strikingly
and significantly exemplified in Japanese films of the silent
era's last decade by the practice of super-inscription, which
seems to have spread with the development of titling itself. 7
It became common for the so-called spoken titles to overlap
with the image which preceded and/or followed them, and
for explanatory titles to overlap with establishing shots at
the beginning of certain sequences. In Makino's 1928
Chitshingura, when a bonze-in-waiting (chabosu) warns
Lord Asano that the all-important ceremony has begun
without him, an allusive cutaway to the ceremony in pro-
gress serves as background for the titles conveying the
bonze's speech. Later, immediately after the assault on
Kira, a 'voice' calls the bonzes to assemble in titles which
appear in the centre of a long shot of the scene and which,
by an animation effect, increase in size as they seem to
7. Super-inscription was used in only a
very few European experiments, such as advance toward the spectator. This highly emphatic device
L'Homme du large, partly because - and it may be observed that this super-inscription does
Western films, unlike those of Japan,
were to be shown to audiences in many often have a dramatic function - produces an effect of
different countries. This consideration volume, but it is on the nearside ofthescreen, as it were, in a
does not diminish the positive
significance of this remarkably space which is not that of the diegesis.
widespread practice in Japan. In Inagaki's The Mother He Never Knew (Mabuta no

120
haha, 1931 ), the hero's challenging shout uttered during the SURFACE AND DEPTH
rescue of an ally 'echoes across the valley' in the form of an
exclamatory group of characters super-inscribed on the pan
shot of a landscape inserted into the editing scheme. In
Fuyushima Taizo's Double Suicide on Mt Toribe
(Toribeyama shinju, 1928), the cry for help of a damsel tied
up in a closet is super-inscribed on a fragment of the design
painted on the closet door, offering a two-fold emblem of
surface.
One of the loveliest examples of this practice is found in a
very late silent film by a famous director of the following
decades, Gosha Heinosuke, one of the Japanese to master
the Western codes of editing and to exploit them unstint-
ingly throughout his career (within a narrative framework
correctly regarded as uniquely Japanese, however).
Gosha's The Dancing Girl of Izu (Izu no odoriko, 1933)
tells of the travels of a group of entertainers making their
way to Izu, the traditional island gathering place of their
trade. A medium long-shot showing the group's arrival in
the village is overlaid by an extensive, extra-diegetic,
'documentary' title giving particulars of the tradition which
brings them here. Later, as the brother of the female lead
holds her in his arms to comfort her over the loss of the
student who had been sharing their travels but has now
gone his own way, the embracing figures are shown framed
in a doorway at the far end of a corridor, and the image is
overlaid by a tanka related to the film's theme and mood.
This composite image, in which one of the depth-producing
strategies, by then commonplace in the Japanese cinema, is
combined with that of super-inscription, is typical of the
continuing acknowledgement of surface, however inciden-
tally, in Japanese film. It is also worth stressing the fac·t that
these extra-diegetic super-inscriptions, all the more strik-
ing as they are the only ones in the film, neither open nor
conclude it. In the sound film of the West, super-inscription
has long been tolerated in credit titles, scroll prologues and
the like, as an intermediate stage between 'reality' and full
absorption in the diegesis; the diegetic effect becomes fully
operative when such titles end and ceases to be so when
they return. In the Japanese view, the diegesis was not
taboo; already the benshi was 'writing on it'.
Even today, this relative indifference towards one of the
vital premises of Western illusionism is instanced daily on
commercial television channels, since at any moment, par-
ticularly during peak viewings, in the midst of the most
thoroughly encoded gangster serial or samurai drama,
large, often ingeniously animated characters may invade
the lower portion of the screen to remind us of the merits of
a deodorant or a Korean barbecue. The procedure of

I2I
CROSS-CURRENTS super-inscription, which has many precedents in Sino-
Japanese culture, from the sutra-inscribed mirrors of the
Fujiwara period to the 'painting-poems' of later eras,
provides us with a remarkably enduring example of the
manner in which the notions of surface and writing reveal
essential affinites within the text-which-is-Japan.

I22
13. KinugasaTeinosuke

Nearly all of the films made during the 1920s by those

who were to become the major figures of the 1930s -
Ozu, Mizoguchi, Shimizu - must unfortunately be
regarded as lost. Knowledge of their early work might
well inflect this attempt to reconstruct the tableau of the
representational modes of the period as determined by
the conflicting currents previously described. To what
extent, we can only surmise on the basis of the earliest
known films (1929 and after). These indicate that Ozu
and Mizoguchi had gone further than most of their con-
temporaries in assimilating the Western mode and codes,
and articulating them in modern subjects. Their lost films
might fill a peripheral area of the tableau which remains
blank. In them, imported codes were probably exploited
for their stylistic potential, as in the work of Murnau or
Sjostrom at their average best.
From all available evidence, however, it seems that
only one director active in the 1920s assimilated the
Western mode (according to the norms of the most
highly Westernized period or modern dramas) and went
on, albeit briefly, to work beyond it as in the advanced
cinema of Europe. There the mode was being stretched
to its breaking point, 'undermined' by the introduction of
basic ambiguities, ultimately 'de-constructed' systemati-
cally (Vertov, Dreyer) by the inscription of its basic
mechanisms within the film, so that one could read the
filmic text relative to the mode as such, considered in its
historical dimension, and not merely through it, as if it
were a 'natural language'. The only director who
assumed this task, utterly thankless in Japan at that time,
was Kinugasa Teinosuke.
Of course, the socio-economic and political circum-
stances which no doubt determined the work of the first
French avant-gardes, the 'Expressionist' /Kammerspiel
movement in Germany (not to mention the more radical
gestures accomplished marginally in both countries in
connection with dada, surrealism and the modernist

123
CROSS-CURRENTS movement in general) had, I have indicated, echoes in
Japan. Strikes, inflation, the rise of a radical working-
class movement (the clandestine Japanese Communist
Party was founded in 1921), attested to Japan's develop-
ment from agrarian feudalism to a form of Imperialist
Monopoly Capitalism in only half a century. The fate of
her economy was already inextricably bound up with that
of the capitalist powers. Yet, of course, the non-
conformity characteristic of important sectors of the
capitalist cinemas of France and Germany was further
determined by super-structural phenomena, i.e. ideologi-
cal or 'cultural' factors. The critical undertakings initiated
by Gance, L'Herbier and Epstein, by Mayer/Wiene and
Robison (and by Man Ray, Leger and Richter) may be
characterized in the larger context of art history, by
reference to the change which took place during the first
quarter of this century and which is described in the fol-
lqwing terms by Gaetan Picon.
Prior to modern art, a work appears as an expression of previ-
ous experience ... the work tells what was conceived or seen,
so that between the experience and the work there is only the
application of a technique of execution. In modern art, the
work is not expression, but creation: it bodies forth something
which had never been seen before, it is formation rather than
reflection. 1
We will recognize in this view the representational/
presentational contrast between kabuki and the Western
theatre as described by Earle Ernst and which is typical
of the antinomy of Japanese and Western arts in general.
However, we must also recognize that today, in the West,
more than ever before it contains the key to what is most
fundamental in the distinction between cinema as cul-
tural activity and cinema as profit-earning commodity.
Given these premises, it might seem legitimate to
deduce an elegant symmetry, whereby Western directors
applied themselves to a critique of the codes of Holly-
wood illusionism - associated or not with a critique of
the existing social order - while Japanese demolished
r. Picon, Gaetan: L' Usage de Ia lecture,
with equal gusto the 'codes', derived from traditional arts
Vol. 2, p. 289. and attitudes, which had achieved undeniable stability by
1920. While the Western avant-garde, in this symmetrical
2. The social cinema of Western
Europe, with very few exceptions (Phil scheme, would be flirting with Eastern presentationalism,
Jutzi, Vigo) took over the codes as they
were, regarding them as a natural
a Japanese elite, seeking to formulate a more-or-less
language, 'innocent' of an ideology virulent social critique, would be assimilating Western
which was regarded in turn as pure
signified, totally enclosed by thediegetic
representational developments - at best looking to the
vehicle. Already challenged in practice German and French social realists, at worst copying
by cubism, dada, surrealism ... and the
Soviet film-makers, this position has
Hollywood.
become theoretically untenable today. This double profile is correct as far as it goes, but it is

124
symmetrical only because it does not go far enough. SURFACE AND DEPTH
Although Japan's attitudes towards representation are
historically and socially determined, it is very difficult to
speak of such widely shared values simply as part of the
ideology of the dominant class, even though they have
been confiscated in recent years by a certain bourgeois
intelligentsia, at least in their purest forms (traditional
theatre, cinema of the 1930s). A basic, inter-class consen-
sus on these matters had existed since time immemorial,
illustrated by the rapid and direct development of the
aristocratic no from the theatre of travelling players and
kabuki from its early associations with the Edo under-
world. A situation which contrasts sharply of course with
the irreversible social segregation in all the arts which
came about in capitalist society during the eighteenth and
nineteenth centuries.
The political and economic upheavals of the eras of
Meiji and Taisho (1912-26) had begun in some important
respects to disrupt that feudal unity in Japan and to pro-
duce the need for the development of a specifically
middle-class ideology, that of industrial capitalism in its
most general form.
At different moments in the development of class
struggle on the one hand and art on the other in the
West, the working-class movement and its petty bour-
geois allies have been led to appropriate the modes of
representation originally developed by the bourgeoisie to
suit its own needs. In capitalist countries, this often
enough produced films which furthered the progress of
radical ideas, but which very rarely took that critical posi-
tion vis-a-vis the dominant modes which is the hallmark
of a century in which radical art has become the main-
stream of art - and in which the struggle of the avant-
gardes runs parallel to but quite separate from that of the
proletariat. In the first country where the balance of
class-power was reversed, these two struggles met, only
too briefly, to 'produce' Maiakovski, Vertov, Tatlin,
Dovzhenko ...
All of these developments, however, were enmeshed
with the whole history of representation and signifying
practice in the West. Our most radical denials of the
dominant modes can be understood only in relation to
the history of their rise and domination; in short, to their
genealogy.
In the Japan of the 1920s, a few film-makers were bent on
reforming the Japanese cinema. Most represented the
needs of a middle class fighting to achieve dominant status,
though, a few years later and for a brief time, some did
speak for the growing proletariat. And the triumphant

125
CROSS-CURRENTS return to the values offeudal Japan in the films of the 1930s
is, as we shall see, the sign of the temporary defeat of both
these classes.
In view of the objective need of a rising class for greater
realism, in view of the fact that the West, with its much
older tradition of class struggle in an industrial society, had
produced a system of representation which it accepted as
realistic, it was understandable that the advanced Japanese
directors should have adopted that system. They were all
the more drawn to it as it was in every way opposed to a
traditional mode which, however popular, was also inti-
mately bound up with the image of the dominant cast of
'feudal capitalists' who still controlled Japan. Whatever the
objective needs of their class, however, the gulf that had to
be bridged by these film-makers was such that it produced
the displacements and condensations already described.
3. With these two major and two or It would be foolish to expect that the process of assimila-
three minor exceptions, the many films
directed by Kinuga~a before 1945 were
tion of such an alien mode could go hand in hand with a
destroyed by fire in that year. However, practical critique of that mode, to presume, in short, that the
Kinugasa himself feels, that these were
his only two 'avant-garde' productions. major gestures of European film-art could be repeated on
such a fragile basis. We must, however, bear in mind that a
4. Kinugasa has no recollection of
having seen the few advanced European century of radical innovative art in the West has shown that
films shown in Japan during the early the arts of Japan and of the Far East in general can be allies,
1920s (e.g. Caligari, La Roue), yet most
Japanese authorities feel that he must as witness encounters made by such figures as Degas,
have seen and been influenced by them. Debussy, Pound, Eisenstein and Brecht.
On the other hand, Kinugasa's decision
not to take Page of Madness with him on A similar experience may be said to have been granted,
his tour of Europe in 1929, to take only only too briefly, to Kinugasa Teinosuke.
Crossways, a far more academic
'Western-style' film, might indicate that Two significant films by Kinugasa have come down to
he was not, indeed, aware that his us. 3 They prove him to be the one Japanese director to have
masterpiece would most likely have
been hailed as such by critics and an elite made a successful attempt to transform the Hollywood
audience who accepted and admired codes in accordance with what he may have regarded as the
Caligari, La Roue and Potemkin, all of
which had gone much 'further' than advanced European approach. As we shall see, the execu-
Crossways. Perhaps he feared that he tion of this project, whether or not it was consciously influ-
had not done justice to his Western
'models', if such indeed they were. Or enced by European film, 4 was ultimately determined by
perhaps he had not indeed been many native factors. However, at this level of interaction
'influenced'. Perhaps this film is simply
the result of a cultural encounter, in a and mutual transformation, the traditional modes were
particularly sophisticated mind which unrecognizable for an audience that 'lived' in them.
had mastered the codes of the West at
an early date, between these and the Kinugasa Teinosuke startt>d to work in films as an oyama
general potential of Japanese culture in 1918. He began his directing career in 1922. To my
vis-il-vis that of the West.
knowledge, only a twenty-minute fragment of one of his
5· Kawabata is known in the West for his silent chambera- Benten, Apprentice (Benten kozo, 1928)-
later novels, The Snow Country, A
Thousand Cranes and The MasterofGo. can still be seen, and it is indistinguishable from the bulk of
similar productions known to us. Kinugasa, however,
6. ' ... it is difficult for us to render
perfectly the posture of the noble former oyama though he was, apparently played a leading
figures of the Court ... since we do not role in the movement to oust the oyama from the cinema,
have access to them. However, we must
enquire as best we can of their habits of and quickly came to share the views of those who wished to
speech, their manners, and pay Westernize Japanese films. In 1926, he founded his own
attention to the observations of persons
of quality.' Zeami in Siefert, Rene, La independent production company, a rare and difficult ven-
Tradition secrete du No, p. 70. ture in Japan at any time, and equipped his own make-shift

126
CROSS-CURRENTS the Pacific War, independent artists who correspond to the
Western image of the original creative temperament,
Kinugasa was undoubtedly the first of these.
The use. of women rather than oyama in female roles, a
contemporary setting and a subject which the 'poetic real-
ists' of France and Germany would certainly not have
rejected, gave Kinugasa's and Kawabata's project an
up-to-date aspect. They quite naturally tried to complete it
by eliminating the benshi. Their 'modern' aims also deter-
mined their decision to eliminate titles as well, an excep-
tional experiment in a 'Western-style' film for that period.
It is fascinating to observe that in Japan as well, at least
one director of cultural ambition was seeking, with respect
to the presence of inter-titles on the screen, 'to turn back
the clock'. As we have seen, the films of the West, like those
of Japan but not for as long, were originally untitled, with a
'lecturer' filling the role of benshi. I believe it can be shown,
however, that in the one Japanese example of this 'revival'
which is known to us, the result is often diametrically
opposed to that of such European experiments as Der
Letzte Mann, Ueberfall or Menilmontant; despite obvious
important differences, it is ultimately related to certain
aspects of Dovzhenko's Earth or to the most seminal of all
the untitled films of the 1920s, Vertov's The Man with a
Movie Camera.
The opening montage sequence of Page of Madness,
striking for its visual stylization, could easily have graced a
German or French film of the late 1920s. Indeed, I feel that
Kinugasa's film was slightly 'ahead of its time' on the West-
ern scale; the narrative sophistication which this film shares
with The Wind, Spione and L'Herbier's L' Argent (all 1928),
was characteristic of the last two or three years of the silent
era.
rain at night; dissolve to ...
close-up of a window
- a door bangs in the wind
- a car's headlights come towards camera
- the car's tyres pass in close-up
two barred windows in chiaroscuro interior (lightning)
water swirling in gutter
the silhouette of a man steps out of the rainy night into a
pool of light
repeat of previous window-shot; overlay
- water lapping against a wall
- close-up of man's feet climbing rain-swept concrete
stairs
- close-up of single barred window (interior)
- a larger view of the same
- a telegraph pole, scarcely visible in the dark

128
- previously seen stairs, from a different angle with water SURFACE AND DEPTH
pouring down them

Gradually, these and similar elements are blended in rapid

montage and increasingly elaborate overlays, until this
veritable swirl of images dissolves to ...

- a row of water-spouts and a huge, glittering tinsel ball, in

front of this a lovely variety entertainer is dancing in a fancy
costume (the dance is shown in a series of staggered over-
lays); the camera tracks back and bars appear in black
silhouette in the foreground; the screen goes black
- an off-centre iris unmasks first a barred window, then a
barred doorway; beyond the doorway, projected on a wall,
is the shadow of the dancing girl in her stage costume;
dissolve to ...
- close-up of the girl dancing in the rags of her hospital
gown

This mood-setting, 'impressionistic' introduction was, in its

general outlines, already part of the codes of sophisticated
European films (such as those of the great Lupu-Pick or
Delluc). After Murnau, Sjostrom, Lubitsch and Leni had
arrived in America, it made its way into the Hollywood
codes proper.
When we examine the development of this sequence, we
observe, however, a number of singularities profoundly
incompatible with the standard Western approach. 'Nor-
mally', when a man appears on a rainy night after a series of
inanimate details to 'set the scene', it marks the introduc-
tion of the human centre of the diegesis. Here, however, the
man - like the car before him, which brought a similar
suggestion - immediately vanishes, never to return. The
man's disappearance is emphasized by the shot of the
empty stairs. The automobile might be linked to the vehi-
cle, seen several reels later, in which patients are brought to
the asylum, or even to the car which appears in the old
janitor's delirium towards the end of the film. Though we
never see his face, it is just possible that the man in the rain
is the janitor's briefly glimpsed son-in-law or one of the
shadowy doctors. These are merely vague possibilities,
however, and the ambivalence remains intact, as it does not
in the narrative editing of the 'brick-by-brick' Pudovkin
tradition. This 'false start' with its ambivalent echoes later
in the film is fundamentally at odds with the impressionistic
chain which ensures a gradual approach to continuity and
the human centre at the beginning of Sjostrom's The Wind.
Moreover, while this portentous introduction might
seem to indicate that the main narrative line deals with

129
CROSS-CURRENTS whatever drama led to the dancer's internment, this
'beginning' is still another strategy of deception. For the
dancer's past is present again only through a glimpse of a
torn photograph on the wall of her cell and in recalls of her
stage costume, to be dealt with shortly. The very simple
main narrative is also predicated on a past/present relation-
ship which, viewed superficially, seems perfectly linear: an
old man's devotion to his schizophrenic wife has led him to
take menial employment in the asylum where she is
interned; memories of their happy past mingle with scenes
of their present misery, culminating in the husband's pathe-
tic attempt to flee with his wife. However, the principles of
spatia-temporal ambivalence suggested in the compressed
evocation of the dancer's drama- black bars in front of the
night-club stage, the costumed shadow on the cell wall- are
brought into play throughout the film, continually challeng-
ing, on the level of articulation, the inherent linearity of the
coded narrative matrix. Close examination of the complex
articulation of temporal narration in this film reveals pro-
cedures which are fundamentally opposed, for example, to
Kirsanoffs time-lapse sequence of grass growing on a
mother's grave articulated by successive dissolves to signify
that her two daughters have grown up. Nor do the 'memory
shots' function at all like those in, say, Pudovkin's The
Mother.
The image of the dancer's costumed shadow provides an
outstanding example. It may be described as a conjunction
of 'past' and 'present' within a single instant of diegetic
space-time which, to the extent that it is neither 'past' nor
'present', is extra-diegetic .7 Now this conjunction was ulti-
mately assimilated as a flashback signal by the Western
narrative codes in films where a connotation of theatricality
made it possible to exclude ambiguities- Benedek's Death
of a Salesman, Sjoberg's Fraken Julie (both I95I). 8 In such
a context, this particular semantic contradiction became a
coded sign, replacing the iris, the dissolve, the wipe, close-
up with surging music, sound-track wind-down, etc., all of
which, at various periods, have had the same function.
In this passage of Kinugasa's film, on the contrary, the
interpenetrations of past and present do not function on a
coded level only. It is true that the bars which appear in
front of the stage and the shadow on the cell wall are
7. But in no sense is this image ultimately coded to the extent that the one may be read as
extra-narrative. This, I feel is a clear
illustration of the fundamental
'premonition of the future', the other as 'remnant of the
distinction to be made between diegesis past'. However, they can be read as such only from a point
and narrative.
which 'commands a view' of the entire sequence. At the
8. Significantly enough, both these films moment of immediate perception, these messages are
were originally plays, and Arthur
Miller's original text called for this very
uncertain, deferred by ambiguities or misleading indica-
device. tions. They are pinned down only as subsequent images
appear. When first seen as fuzzy, black bands, the bars do SURFACE AND DEPTH
not look like bars at all, and when they become sharp, they
might easily be prison bars. And the possibility exists that
the girl, whom we have not yet seen in her inmate's rags, is
actually wearing her costume in her cell (an image which
actually appears later in the film). These circuitous trans-
missions of what are, at best, messages of low semantic
density -lacking, in particular, any subjective dimension:
memory, here, is anonymous- result in the partial expul-
sion of these images from the diegesis (where instantane-
ous transmission is the rule) and in their rising to the surface
of the film.
The shot of the shadow on the wall bears a significant
relationship to the pivot-word (kakekotoba ), central to
Japanese traditional poetic forms; like the pivot-word, it is
the point of junction of two separate systems which 'over-
lap' at that point, and it functions simultaneously as a part
of each system. In Page of Madness this procedure never
becomes a coded 'punctuation', for though it is used twice
more, each time it assumes different forms with different
results.
The girl dancing in her cell is a theme that frequently
recurs, but only twice again does the costumed figure from
the girl's 'past' appear in her drab present: once 'in the
eyes' of the male inmates, frantic with sexual excitement, as
they crowd against her cell door to watch her dance; and
once in conjunction with the old man's memories of a
festival attended with his wife and daughter. In the latter
sequence, the image is contrasted with a 'reality' in which
she is not dancing but lying bound hand and foot on the
floor of her cell, following the near-riot caused by her
exhibition; in the former, it works against shots of her
dancing in her inmate's garb. Here too we encounter
ambiguities and deferments, but the itineraries are very
different. The first shot of the costumed girl dancing in her
cell has optical distortion and is preceded by a track-in on
an inmate's face pressed to the bar: at that moment it
appears as a libidinous, first-person fantasy vision, yet at the
same time it is an anonymous memory of the original
night-club shot in the prologue and of the 'pivot-shot'
which introduced the scenes of life in the asylum. This shot
too is a pivot-shot but the over-lapping systems are of a
different order and are related in a different way. The
second appearance of this shot brings a third type of pivot
structure: when it first appears, it 'rhymes' by analogy with
the father's recollections of the festival and his happiness
with wife and family. When it is followed by a shot of the
bound dancer on the floor, it acquires, for the first and only
time, the cast of subjective memory.

I3I
CROSS-CURRENTS We are dealing with an approach to semiosis not unre-
lated to that encountered in Tanaka's The Red Bat, but of a
much more sophisticated kind. The patently random,
unorganized circulation of free-floating signifiers charac-
teristic of Tanaka's (and Ito's) films-and more generally of
a particular cultural filter- is replaced, in Page of Madness,
by a systematic exploration of the combinations offered by
similar types of polyvalence. It is in this sense that we may
speak of Kinugasa's film as containing play (i.e. creative
work) on the codes directly related to the masterworks of
Vertov and Dovzhenko cited above.
The film does not, of course, consistently maintain this
level of complex tension with respect to the narrative
codes. The shots of mental evocation- and there are many
- are often unequivocally subjective, such as that of the
insane wife's face as a young woman, seen in a mirror, a
shot which is cut into a sequence of demented primping.
However, Kinugasa's preoccupation with conjoining the
de-constructing and structuring processes, still extremely
rare at that period even in the Soviet Union, is evident in
many other instances.
Three times in the film the old man looks out of the
window of his shabby room at the asylum. In two of these
episodes, what he 'sees' is ultimately designated as his
'memory': the festival referred to above (confused with an
'actual' festival to such an extent that it takes many view-
ings to discover the 'seam' between them) and his wife's
dramatic arrival at the asylum. The first of these 'flash-
backs' is presented in a 'deferred' version of a standard
form - a shot of the old man looking out of the window is
followed by the shots of the daylight ('actual') festival
which will lead to the nocturnal ('past') one. The wife's
arrival has an even more ambiguous status, since it is pre-
ceded by a shot in which the old man leaves his room, and
since the association looking-out-of-window/old-man's-
memories is confirmed only in retrospect (a shot of the old
man at the window immediately follows the arrival scene).
By now it seems established that the association is a coded
one: what the old man sees out of the window are, ulti-
mately, his memories. Whence the ambiguity of the third
episode in this series, since this time he looks down on what
we can only perceive as an 'actuality': children mimicking
the behaviour of the patients. And while it is precisely the
theatrical nature of this scene which is the key to its 'real-
ity', at the same time this theatricality relates it to the
fantasy-exploration of the film as a whole.
'Acted out' fantasies- seldom those of the patients, most
frequently those of the old man - occupy an important
place in the film. However, while the climactic hallucina-

132
tions- in which the old man sees his daughter in wedding SURFACE AND DEPTH
dress, presumably driving off on her honeymoon with the
bearded inmate who lusts after her- is a visually impressive
bit of bravura, its 'surrealist' mentalism is perfectly coded.
A European reference would be Mosjoukine's Le Brasier
ardent (1923), with which this sequence compares favour-
ably. On the other hand there are many less spectacular
sequences which offer remarkable non-linear structures,
related to those already described.
Early in the film, at the end of the first dancing episode
referred to above, the old man appears on the screen for the
first time, coming to visit his poor mad wife. This section,
too, is worth describing in some detail.

- repetition of the earlier shot of barred windows back-lit

by lightning; the old man makes his initial entrance, in a
knee-length shot, silhouetted against the windows
- the girl rises from her prone position on the floor and
resumes her dancing
- the old woman is lying on her stomach, seemingly watch-
ing the dance ('Kuleshov effect': 9 the women are in sepa-
rate cells and cannot see each other)
- another shot of the old woman from a different angle
- a titled shot of the grill; the old man enters in overlay
wearing a 'splendid' uniform; he puts his hand on the grill
which opens in the overlay but remains closed in the original
image
- the woman looks up to her left and speaks to her husband

There ensues a correctly matched reverse-field series as

they exchange a few words. Then the woman gazes around
her.

- image of a baby crying overlaid by that of a ventilation

grill
- close-up of the woman; she seems to have a moment of
absence, then looks up again, but this time to the right
- the original shot of the barred doorway: there is no one
there
- the woman turns her head slowly to the right again, then
drops it to her chest
- in the empty cell corridor, the camera tracks towards the
large grill which closes off the end; a very slow dissolve (in
fact, for a time, an overlay) combines this forward move-
ment with a backward movement in the same set so that, at
the end of the shot, the camera is in its original position; the
old man comes down the corridor peering into the cells g. Imaginary spatio·temporal
juxtaposition through 'deceptive' use of
(Fig. 5) matching procedures.

133
past station in life. The ventilator-over-baby shot brings1n SURFACE AND DEPTH
the presence/absence dichotomy, and for this very reason
constitutes a momentary suspension of the fantasy, even
though, from a linear viewpoint, we are still 'in it'. The
dissolve on the empty grill clearly marks the 'return to
reality', thus giving possible meaning to the disconcerting
eyeline match that marked the start of the encounter. As
for the double dolly movement in the corridor, its remark-
able ambiguity comes from being 'superfluous': it is the
second or the fourth time (counting the baby and the mis-
match) that the return to reality has been announced, and
here it is done with such extravagance (the movements are
slow and the shot is long) that for a while we seem to have
returned to the system of pure surfaces which characterized
much of the earlier imagery.
This sequence as a whole is made more ambiguous by the
fact that while the fantasy of the old man's uniformed
appearance at the grill is ultimately designated as the
woman's fantasy, its appearance on the screen is also
determined by the husband's 'actual' intention to visit her, as
conveyed by the shot of him passing the windows. This situa-
tion precludes any simple mentalist reading of the fantasy
and its indices; both it and they are over-determined.
This non-linear approach to the signification process is
not confined to fantasy sequences. One of its more remark-
able instances is the sequence that follows the daughter's
departure at the end of her first visit and the ensuing
near-riot caused by the dancer.

- the old man has been called into the office of t;e head-
doctor, who rebukes him for his involvement in the riot;
dissolve to ...
- close-up of a spinning railway-ticket rack; overlaid on
this, the daughter in a medium close-up takes a ticket from
one of several such racks
- close-up: she punches the ticket
- she hands the ticket through a window to an invisible
customer; in overlay, the wheels of a train start to turn; very
slow dissolve to a coin lying on the sidewalk; clogged feet
pass
- detail of curtain and wall-paper; truck back to full view
of apartment where the young woman lives with her hus-
band10

Now, Kinugasa made a sound version of the film after his

ro. The rest of this sequence contains an
accidental rediscovery of the negative and a print among ambiguity so startling that one hesitates
his personal belongings in 1971. Among the very few to attempt analysis: the young woman
looks through a door and appears to see
sound-effects added to an almost exclusively musical track herself talking to her husband, who is
were a train-whistle over the close-up of the spinning ticket lying on a sofa.

135
CROSS-CURRENTS rack and the sound of a train gathering speed over the
ensuing images of the ticket office and the wheels turning.
These are clearly intended as reinforcements of temporal
linearity and to re-establish semantic 'clarity'. The whistle
helps to pin down the visually ambiguous image of the
ticket rack, while the ambiguity of the following images (is
that train merely an associative image, or does the girl take
it to go home; and what, if any, is the 'meaning' of that coin
on the sidewalk?) is overridden, so to speak, by the impetus
of the train sound. Posthumous correction of early films in
this way often points to what is still today most subversive
about them.
In Japan, sophisticated freedom with the codes of narra-
Live linearity disappears until the late 1950s and early 1960s.
It is all the more remarkable that the structures of such
'ldvanced Soviet films as The General Line, Earth or The
Man with a Movie Camera, all of which are slightly later
than Page of Madness, may be considered to be prefigured
in this remarkable film.
Kinugasa, however, was discouraged by the hostility of
nearly all the critics and by the film's commercial failure.
His next independent venture, Crossways (Jujiro, 1928),
was far more faithful to the Hollywood and European
codes. Although the film's imagery has been compared
with Japanese traditional ink-painting (sumi-e ), it is far
Figs. 6, 7· Kinugasa more obviously related to that oflatter-day Expressionism/
Teinosuke, Crossways Kammerspiel. One long and very impressive sequence
does, however, recall the complexities of Page of Madness:
the hero, a poor ronin who has been temporarily blinded in
a quarrel at the Yoshiwara over a famous courtesan whom
he loves, has feverish visions of her (Fig. 8) and of the gaudy
revelry of that famous 'entertainment quarter'. His fan-
tasies are often joinedseamlessly with sequences which are,
in fact, not 'visions' at all but parallel action ('While our
hero is raving, this is what is happening at the Yoshiwara').
In this film, the hallucinations, like those in the final sequ-
ence of Mabuse or the vision of Moloch in Metropolis,
affect at times the material appearance of the set, as in the
scene in which the hero's poor hovel becomes a field
planted with jars containing boiling water, which burns his
parched throat (Fig. 9).
The mastery with which such structures are manipulated
in these sections of Crossways is equal to that displayed
throughout Page of Madness. However, in confining this
type of ambiguous articulation to passages designated as
hallucinatory, Kinugasa was making the decisive concession
to codicity which ultimately brings this film into line with the
psychologism of the dominant Western cinema. Any viola-
tion of the principles of illusionism must automatically
'" ~-:0!::. :f::' -F)i:--'' J~~r::;ne:·;;:' -~j;·;"" '~~-<·
<' ::---~~ .!'
' >,.,_ '-\ -~·:;.; ;:-~~' ,,;_ u): :::.n::;:;::. :lr;(:. :J·-.~;; t·,
ti :a.L :;:- _-.i ~nda~'y. ~-: :1~; ~~ ,., ;~:;;;:i.:"""rl~~J;;:·
J ; ~:!i~"i:? i">-'L.·p;:_ }f ~-<:~:~ H1
Part4 Iron Trees
Golden Flowers
14. The Weight of History
and Technology

The two major traumas of the post-war period had been

inflation and the Kanto earthquake of 1923, which
shattered much of the country's industrial structure. Con-
jointly, they produced two complementary and not unex-
pected results. On the one hand, class struggle began to
develop rapidly, with Marxist ideas penetrating the urban
proletariat and certain fringes of a liberal bourgeoisie
which, by the late 1920s, had acquired some political
weight. The political parties, led largely by middle-class
civilians, had succeeded in creating a new balance of
power and were pressing for an even more complete
'Westernization' of Japanese political and social norms,
indispensable, they realized, to the 'healthy' growth of
capitalism. On the other hand, there was a spectacular
rise in the activity of the so-called 'patriotic societies',
radical right-wing clubs some of which had existed since
Meiji times, all of which now displayed an unprecedented
virulence. The conditions that gave rise to these societies
and the motives of the men who formed or radicalized
them have been summed up thus by an American his-
torian.

The growth of industry . . . was helping to undermine the

attitudes proper to a disciplined and dedicated people, for its
profits tempted those who shared them into new forms of
extravagance, nearly all imported from the West, while their
unequal distribution led to 'dangerous thoughts', also of West-
ern origin, like socialism, pacifism and democracy. Similarly,
modernization of the economy could be blamed for rural dis-
tress and hence for weakening the position of the farmer, soci-
ety's staunchest upholder of traditional behaviour. In fact,
dance halls, luxury, political corruption, big business, trade
unions, strikes, agrarian unrest and debased standards of every
kind, all could be lumped together as results of an over-
indulgence in foreign ways. They thus became a focus for the
resentments of men of many different kinds: those who felt
that the new order of things gave them less than their proper

141
IRON TREES, station; those who genuinely respected the past and the values it
GOLDEN FLOWERS represented; and those whose sense of inferiority in the face of
the West's achievements brought a hatred of factories, as well
as ambition for empire. The resulting movement embraced con-
servatives, professional patriots, agrarian idealists, advocates of
state ownership and social revolutionaries, all contributing in
some measure to the aggressive 'ultranationalism' of the
nineteen-thirties. 1

Whatever their diverging social origins, these men were

all bent on reactivating what we clearly recognize as the
official ideology of the Tokugawa bakufu, with its dis-
trust of commerce, industry and science and of all that
was not Japanese. A belated feudal ideology was stub-
bornly determined to stem the tide of history, the
accelerating rise of the bourgeoisie and the development
of proletarian consciousness. The elected enemies of all
these groups were the zaibatsu 2 who continued to deter-
mine the realities of economic power in Japan and who
were increasingly aware of the need to strengthen the
attributes of bourgeois democracy in order to consolidate
their political power. During the early and middle 1930s,
the extremists virtually paralysed government through a
campaign of systematic assassination and intimidation.
Their goals came to coincide with those of a traditional
military cast, who received support from young officers
of peasant origin, embittered by the desperate plight of
the country folk, and who were bent on widening Japan's
foothold on the continent. The Mukden 'incident' and
the annexation of Manchuria in 193 I (followed, signific-
antly, by Japan's withdrawal from the League of Nations
in 1933) coincided with a serious trade slump, due, in
particular, to the Wall Street crash and the collapse of
the world silk market.
From that time on, Japan began, economically and
psychologically, to withdraw into herself, to reject West-
ern ideas and fashions. She began to see these through
the eyes of the patriotic extremists.
Foreign visitors were accused of spying on the flimsiest pre-
texts. There were arguments about the use of foreign words and
whether nameboards at railway stations should read from right
to left (Japanese style) or left to right (Western style). Much of
this chauvinistic atmosphere was injected into the schools and
1. Beasley. w. G .• The Modern History universities. Many foreign books used in them were proscribed
of Japan. P· 2 37· by the police - often without a very clear idea of their contents
2. zaibatsu: the financial and industrial -and text-books were rewritten in nationalist terms. 3
clans, largely of aristocratic origins.
formed during the Meiji era. By the mid- 1930s, the jingoist propaganda had found fer-
3. ibid. pp. 22 5--6. tile soil in a country under economic pressure. Shakai

142
Taishuto remained the only moderately left-wing opposi- THE WEIGHT OF HISTORY
tion party after the physical elimination of the Commun- AND TECHNOLOGY
ist Party and the rest of the radical left by 3 ,ooo arrests in
October of 1932. It made record gains in the elections of
1937, but this reaction on the part of the urban masses,
while attesting to a very real current of dissenting opin-
ion, came too late to prevent the impending tragedy.
I have dwelt at some length on the political and
ideological climate of the period because it is directly
relevant to our subject. Yet this relevance is problematic,
to say the least, since the most fruitful, original period in
the history of a nation's cinema coincided with the rise of
a regime and a national ideology akin to European Fas-
cism. It was between 1934 and 1943 that Ozu and Miz-
oguchi produced their finest work, and that Naruse,
Shimizu, Ishida, Yamanaka and many others helped to
perfect an approach to film-making that was not only
uniquely Japanese but was equal, at its best, to the finest
achievements of the traditional arts of previous centuries.
A comparison with Germany, where Nazism quickly
and effectively strangled anything resembling a living
cinema, is of no value, since Hitler's putsch caused the
wholesale and immediate exodus of both celebrated and
budding talents. Had liberal intellectuals like Ozu, Miz-
oguchi or Ishida ever wanted to leave (and this is quite
unlikely) there was no place for them to go. The linguis-
tic and cultural barriers to departure were insurmount-
able.4 However, most of the major figures of the Italian
industry also went on working in their native country
under Fascism, and the slick, conformist films of Blasetti
and Camerini give certain evidence of the sterilizing
power of Fascism, observable in every other art.
It is here that we observe a significant difference (not,
of course, an intrinsic 'superiority') which sets Japanese
military Fascism apart from its Western equivalent. For
while what we may legitimately regard as the 'golden
age' of Japanese cinema was certainly overdetermined,
one factor is that withdrawal of Japan into herself, the
rejection of Western values by the nation as a whole. The
nationalism of the 1930s was a mass phenomenon to an 4· To my knowledge, no Japanese
extent and in ways which that of the European Fascist director has ever made a career abroad,
as so many Europeans have in America,
countries was not, 5 and at the cultural level it struck par- or a few Americans in Europe. The
ticularly sensitive chords in both the masses and the intel- contemporary director, Hani Susumi,
who has shot several
ligentsia. The relations between film history and the Japanese-produced films in the West, is
political history of that period are of course complex; as as close to an exception as one can find.
we shall see, themes of mild social protest do appear in 5· There was no equivalent in Japan
the work of the masters. One thing, however, is quite either to the physical elimination of an
important part of the German
clear: the social pressure to adopt Western modes undoub- population or. conversely, to the Italian
tedly abated during those dark years. As we have seen, in resistance movement.

143
IRON TREES, so far as cinema was concerned, those modes had never
GOLDEN FLOWERS really taken root in the mass audience. There was ample
room for the growth of a specifically national cinema.
It is to the credit of Mizoguchi, Ozu and most of their
noted contemporaries that the politically nationalistic
film seems to have remained rare in their work until the
end of the 1930s. 6 The maintenance in so many films of
traditional values, in both narrative form and mode of
representation, is nevertheless clearly consonant with the
ideological indoctrination organized by the caste that was
rapidly taking over Japan at that time.
We must not forget that in Japan this concept of a specifi-
cally national cinema takes on a meaning unknown in the
West, since it implied the development of a non- Western,
specifically Asiatic film. And of course it is no accident that
the tacit refutation of the Hollywood codes and of their
claim to universality (as opposed to the 'passive resistance'
of the 1920s) was sustained in the background, so to speak,
by the pan-Asian ideology which the Japanese Right had
gradually been forging to justify their imperialist aims. The
doctrine known as the Greater East Asia Co-Prosperity
Sphere was a deadly threat to China's fragile and still
largely formal democracy and independence; it was also,
nonetheless, a reaction to an equally important historical
reality: the colonial exploitation of Asia by the Western
powers over the past four centuries.
Let us now turn to the situation of the film industry
proper. The major event of the early depression years was
the appearance of talkies from the West. The sound film
produced two fundamental changes everywhere. First, it
immediately showed that it was potentially capable of
greatly broadening the film-going audience. At the same
time, of course, it called for much heavier investments,
ultimately justified by its greater box-office potential, but
which in Japan raised problems less easily solved than in
Europe or the United States. These problems derived
mainly from the small profit-margins of Japanese cinema
and the need of investments from banks and big business, a
practice which was not normal in Japan and which was not
established for many years. The result was a situation which
was practically unique among the developed film industries
of the world. In Japan, silent films were still being produced
as late as 1937, and in 1934 they still represented the bulk of
production.
6. However, Mizoguchi'searly film, The In Western Europe and the United States, on the other
Dawn of the Founding of Manchukuo
and Mongolia offers a striking contrast
hand, they had disappeared by 1931, in the Soviet Union
with the 'tendency' films that had by 1934. This technical advance had a fundamental effect
preceded it, a contrast emblematic of
the ambivalent status of cinema in Japan
on the composition of the film-going audience of the West.
of the 1 930s. In France, polls taken by trade papers showed that in 1927,

144
over three-fourths of the population never went to the THE WEIGHT OF HISTORY
cinema at all. And it is common knowledge that the French AND TECHNOLOGY
audience was overwhelmingly urban and proletarian - a
situation which was reflected to a degree in all developed
capitalist societies. It was not until the coming of sound that
the bourgeoisie began to attend films en masse, for the
obvious reason that films had at last begun to talk. Edison's
project was fulfilled, films had at last attained the status of
the ultimate performing art, the 'realistic' theatre. And, of
course, the aspects of cinema that did not meet with this
expectation were rejected. The pressure to produce
'canned theatre' for the new spectators, who could pay more
and attend more frequently, was tremendous; few directors
were in a position to resist. The 'poetic licence' of the silent
film, its non-realist stylizations, its lack of the Word, had
become acceptable to a predominantly proletarian
audience, not yet thoroughly formed by an ideology of
transparent representation, simply as a result of cultural
segregation, still so complete at that time. Reflecting, no
doubt, this 'open-mindedness', even the hack critics of the
day seem to have taken that freedom in their stride (witness
their calm reception of bold works, such as Caligari,
Potemkin and the French 'Impressionists'). It was, how-
ever, to be brutally eliminated from the professional
cinema of the United States and Europe for some thirty
years. As we view in retrospect the ostracism that stifled
Dreyer after Vampyr, Stroheim after Queen Kelly, and
brought about the more or less involuntary resignation of
such major figures as Gance, L'Herbier, Dulac, Epstein
and perhaps above all, Fritz Lang, in work of desperate
conformity, we see that the late 1920s and early 1930s
constituted the final and most spectacular stage in the
saturation of Western cinema by the modes of representa-
tion of the dominant class.
Paradoxically, it was during the 1930s that the Japanese
cinema, wholly out-of-phase with this Western pattern,
'comes into its own'. Three major factors contribute to this
apparent anomaly. One has already been discussed: the
'return to Japanese values' which marked the political and
social scene. Secondly, the bourgeoisie which indeed began
to attend the cinema more frequently as films began to talk,
remained a Japanese bourgeoisie, with its heritage of tradi-
tional culture. We may compare their need for 'realism' to
that of their forebears under the Tokugawa bakufu. It still
involved attitudes unlike those which we associate with the
bourgeois ideology of representation in the West. Thirdly,
there was the economic and technological factor, i.e. the
decade required for the spread of sound.
According to the novelist Kishida Kunio/ the first 7· In Japan Film Yearbook, '937·

145
IRON TREES, Japanese sound film was Osanai Kaoru's Reimai (Dawn,
GOLDEN FLOWERS 1926?), a faithful rendering of a shin-geki production.
Osanai, the co-director with Murata of Souls on the Road,
was primarily a man of the theatre. His film was presum-
ably shot by the De Forest Phonofilm process, which had
been demonstrated in Japan the year before. However, as I
have said, the talkie did not completely supplant the silent
photoplay on Japanese screens for more than a dozen
years, and nothing resembling the regular production of
sound films was begun until 1932. The first commercially
and 'artistically' successful venture, all authorities seem to
agree, was Gosha's The Neighbour's Wife and Mine
(Madamu to nyobo), made in that year.
This extremely gradual introduction of sound is interest-
ing in a number of ways. The longest period of resistance to
sound occurs at both ends of the cultural scale. Ozu and
Naruse did not make their first sound films until 1936.
Gosho, after the film cited above, returned to silent pro-
duction until 1935. Mizoguchi, after 1930, alternated sound
and silent films, the last of which dates from 1935 also. So
that on the one hand, silence- and the benshi- still strongly
tempted well-established film-makers, who already had a
cultural reputation. Their attitude might seem to reflect
that of the many Western directors who resisted sound to
the point of temporary retirement. Consider, however, that
in 1936, when 259 regularly operating theatres out of a total
of r ,627 in Japan proper were still showing only silent films,
at least one company, the Daito Motion Picture Co. Ltd, a
subsidiary of Shochiku, was producing only silent films
8. ibid., Ichikawa Sai, 'An Outlook of
(this was still the case the following year); and that in that
Motion Picture Industry in Japan' [sic]. year it produced ror films out of a national total of 531!
The figures cited here are from this
source. Actually, the company may well
Moreover, and most important, these ror films were 'prin-
have been the Shink6 Kinema Co. Ltd. cipally cheap, second-rate pictures' .8
another Shochiku subsidiary which
produced 99 films that year (the doubt
In other words, the resistance of the most sophisticated
arising from a regrettable tendency film-makers to the coming of sound also corresponded to
among the translators of that otherwise
invaluable Yearbook to confuse 'latter'
the surviving popularity of the silent film. Whatever the
and 'former'). The Yearbook's ideological reasons for the former and whatever the socio-
publication in 1937 for the first time is
significant of the fact that, as Ichikawa
economic reasons for the latter, both facts, clearly over-
points out, 'the experimental period for determined, constitute a single complex historical situation
talkies [was] at last terminated' and 'a
remarkable tendency of businessmen to
of considerable significance. They also clarify the lingering
invest capital in the motion picture survival of the benshi, who apparently did not vanish com-
business [was ]looked upon as a
foreboding of dawn upon the motion
pletely from commercial theatres until shortly before the
picture industry.' Unfortunately only Pacific War. The benshi played an active role as a pressure
one further issue of the Yearbook
appeared in the following year. It
group in the persistence of the silent film. The 1932 strike of
presumably succumbed to the lack of Nikkatsu benshi is a celebrated episode. However, it is also
interest abroad in Japanese productions
and to the country's developing war
clear that while film-makers like Ozu and Mizoguchi were
economy. (Rare copies of both issues are unlikely to have been outspoken partisans of the benshi,
preserved at the BFI Library (London)
and the Cinematheque Royale de
they accepted his virtually inevitable presence beneath the
Belgique.) silent screen rather than convert to sound. In other words,
they made use of the benshi and his distancing role during THE WEIGHT OF HISTORY
the formative period of their mature styles. They post- AND TECHNOLOGY
poned coming to grips with sound until they felt capable of
maintaining, in and through their mature styles, a 'tradi-
tional' distance with respect to Western representational-
ism as reinforced by synchronous sound.
This hypothetical explanation suggests another: if any
period of European film may be regarded as its Golden
Age, it was the five years between the first commercializa-
tion of synch sound recording in 1927 and the ultimate,
stultifying spread of 'canned theatre' under the combined
pressure of German and American capital via Tobis and
Western Electric, in 1932. Has any other comparable
period of time seen such an extraordinarily rich harvest? It
includes Epstein's La Glace atrois faces and L' Or des mers,
Gance's Napoleon, L'Herbier's L'Argent, Renoir's La
Chienne and La Nuit du carrefour, the complete works of
Jean Vigo, Bufiuel's Un Chien andalou and L'Age d'or,
Cocteau's Le Sang d'un poete, Dreyer's La Passion de
Jeanne d'Arc and Vampyr, Lang's Spione, M, and Das
Testament von Doktor Mabuse, Sternberg's Der Blaue
Engel, Dudow and Brecht's Kuhle Wampe, Vertov's Man
with a Movie Camera, Eisenstein's October and The Gen-
eral Line, Dovzhenko's Earth and Arsenal, Barnett's The
Thaw. 9 Even such minor figures as Pabst (Westfront I9I8,
Kameradschaft) or Kuleshov (The Great Consoler) pro-
duced at that period what were far and away their crowning
achievements. Not only were the silent films named here
among the ultimate attainments of the silent period, but
these very early sound films remain to this day among the
supreme examples of the 'creative use' of sound, thanks to
the three-year delay before the triumph of 'canned theatre'
with the absolute priority it gave to dialogue and its increas-
ingly pervasive back-drop of 'movie music'. During the first
few years of the sound period in the West, a relatively large
number of film-makers were able to extend their experi-
ments with images and graphic signs to the sound-stage. It
seems reasonable to suggest that in a country where this 9· In the Soviet Union, too, sound
transitional period lasted over ten years, and despite our ultimately brought a complete change in
the scope of cinematic creation, partly for
relative ignorance of the films made during this period, that reasons related to Western
the interregnum had a similarly stimulating effect in Japan. developments. The political cadres
suddenly found themselves confronted
There is one fundamental difference between the with a medium which involved spoken
achievements of the great Japanese directors of the 1930s discourse, i.e. the very substance of their
political practice. Their competence to
and the great Western directors of the 1920s. As I hope to intervene far more directly than before
demonstrate at length, this was a period in which the was suddenly legitimized. The results are
only too familiar. In particular, the
aesthetic values of Japan's past came to be fully reincar- decision to replace 'typage' and
nated in cinema. We have observed that originality was the mass-hero by a star system and the
positive socialist hero is curiously parallel
never a primary value in Japan, particularly in art. The to similar developments in Western
individual artist's contribution was not evaluated, by himself Europe and the United States.

147
IRON TREES, or by his society, in terms of his aptitude for inventing
GOLDEN FLOWERS new forms, or the 'revolutionary' quality of his work. On
the contrary, the subtlety with which he recombined ele-
ments that had served for generations, the way in which he
introduced cunning changes, fresh but slight, into the work
of his predecessors, was valued above all else. Originality,
as such, was inconceivable except when fused with evident
mastery of the methods and spirit of tradition. Such was the
character of Japan's cinema during its first half-century.
During the 1930s and 1940s, this character was developed
to its highest degree.
Supreme as the individual achievements of Ozu and
Mizoguchi may appear to us today, there is no break in
continuity between their work and that of a large body of
directors working in similar genres. Despite the ready
adoption of the Hollywood codes of decoupage by Gosho
and Shimazu, for example, their work, considered collec-
tively, is nearly as remote from that of the dominant cinema
of Europe and America as that of Ozu, Mizoguchi and
Shimizu, whom J view as no more nor less than the supreme
masters of a unified cultural practice. They are to the
cinema of that period what Ki no Tsurayuki was to Heian
poetry, Sotatsu and Korin to seventeenth-century screen-
painting, Zeami to the no or Chikamatsu to the doll
theatre. None of these can be said to have towered over
their period or their art as Shakespeare, Beethoven or
Cezanne did.
Although this fundamental distinction may be chal-
lenged by Western (and Japanese) adepts of the 'auteur
theory', at least one Japanese observer of the period was
fully sensitive to it. Katumoto Seiitiro was clearly aware of
the profound community among Japanese directors of the
period, of the importance for them of the Japanese past,
and of the deep gulf that separated the cinema of Japan
from that of the West. In a remarkable article, he not only
recognizes but briefly describes the Japanese mode of
filmic representation.

The attitude of these film directors is something like the mental

position assumed by Japanese poets of the tanka and haiku
schools since ancient times. Not all of them, of course, are con-
scious of the notion that their own particular art might have a close
connection with the traditional poetry of their country. They
evidently believe that they are striding youthfully along the path
of modern art [i.e. Western art]. Judged dispassionately and
objectively, however, there is a surprisingly good deal of the
ancient Japanese poet in them, their modern appearance to the
contrary notwithstanding ...
These historical circumstances have also affected the Japanese
cinema today. Our film directors and actors have, of course, seen THE WEIGHT OF HISTORY,
American and European pictures here in Japan and been influ- AND TECHNOLOGY
enced by them. They have even employed certain effects that
outwardly appear to be similar to those of the foreign film.
Nevertheless, the tradition back of the technique that has been
followed in this country and, accordingly, the innermost artistic
sense, are entirely at variance with those of Europe and America.
This difference is clearly revealed everywhere in the general run
of pictures produced in Japan ...
. . . the Japanese cinema is no other than a means which pro-
vides images to the story-telling performance of the feudal age
and the popular novels of today. A continuity ofillustrations as it
were, that revolve on an axis consisting of a typical narrative tale-
such is the essential nature of the Japanese cinema today.
Obviously the constructive method of the Japanese cinema
which has pursued this narrative medium is a somewhat different
thing from the so-called 'pure movie constructive sense', such as is
employed by American and European directors and which is
considerably removed from the [traditional Japanese) literary
narrative method. Whether or not the difference between the two
can be explained away by the simple statement that the American
and European schools constitute an advanced step in the making
of film plays, and that the Japanese school is a step behind the
times, is a moot question indeed. The dual aspect of the Japanese
cinema, namely that on the one hand it has as its central factor the
constructive method as indicated above and that, on the other
hand, it is conceived along lyrical and suggestive lines, is fostered
by the principle of reciprocal causality. Moreover, they are both
characteristics which will cling tenaciously to the core of the film
plays of this country, and whether our directors will in future
attempt to elevate the quality of their pictures by cutting loose
from those fetters, or whether they will strive all the more to
develop these features and attain a higher level- this, indeed, is a
serious problem which is pregnant with interest. 10

IO. Katumoto Seiitiro, 'Characteristics of

To complete this contemporary corroboration of views the Japanese Cinema', in]apan Film
which I will develop in the following chapters, a few words Yearbook, 1936. The italics are mine.
are in order about the difference, also pin-pointed by I I . This may seem contradictory in view
Katumoto, between the development of cinema and of my claim that the development of
Western cinema after I 909 was not
theatre in Japan, on the one hand, and music, painting and simply a response to the tastes of a mass
literature on the other. I actually know little of the social audience but, ultimately, an imposition
upon that mass audience ofthe tastes and
place of shin-geki, but it seems quite clear that unlike such conceptions of the ruling classes. This
resolutely elitist, neo-Western forms as the 'serious novel', contradiction does not obtain, however,
given the relative unanimity of taste in
classical music and oil-painting, which developed in an Japan until very recent times. The
isolated, almost artificial world, cut off from any mass cinema of the 193os reflected tastes which
were shared more or less equally by the
audience, the cinema was, from the very beginning, a popu- producing and consuming classes. Or at
lar art, appealing to the concrete reality of Japanese taste, 11 least they were on the most basic level of
representation. We cannot assume that all
formed by a unique historical experience. films appealed equally to all classes,
All the contributors to the 1937 Yearbook concur in regardless of genre and subject-matter.

149
IRON TREES, substance with Uchida Kisao that 'the minority of the
GOLDEN FLOWERS public who prefer foreign pictures to Japanese ones are of a
higher intellectual level than the majority who are satisfied
with pictures made in Japan. Indeed, some critics have
gone so far as to state that the former form a class all by
themselves. Such being the case, one may well understand
12. Uchida Kisao, 'The Motion Picture
why it is there are to be found in every important city in
Theatres in ]apan',]apan Film Japan theatres which make it a business of screening only
Yearbook, 1937, p. 53· Several other
contributors stress the fact that a
imported films' .12 And it was of course the same elite which
movement towards mixed programmes was reading Western or Western-style literature, listening
had begun (presumably because of a
scarcity of native sound films). Today the
to Western or Western-style music. However, a very large
segregation principle is completely majority in every walk of life were watching Japanese films.
restored in the cities; some theatres have
two or more halls, one of them reserved
And until the mid-r930s a majority of these films were 'all
for foreign films. the more Japanese' as they were silent.
15. Some Remarks on the
Genre Syndrome

Before turning to the cinema of Ozu, so firmly anchored

in a single genre, we must consider one important gen-
eral aspect of the Japanese attitude towards cinema.
Although present since the beginning, its relevance to
our theoretically oriented investigation begins in the
1930s. As all observers of Japanese cinema have pointed
out, numerous strict genre classifications were set up and
maintained at all costs, in production and programming
as well as in publicity and criticism. The distinction between
jidai-geki and gendai-geki (period as against modern
'plays') may appear identical with the genre distinctions of
the West, costume-drama, gangster film, etc. However,
there are, I believe, two important differences.
Japanese genre taxonomy has been developed and
used, not to establish hierarchies of differentiation, as so
often in the West, but in order to emphasize an objective
unity/plurality relationship. This is achieved partly
through linguistic effect, by the repetition of a root word
(though an occasional departure from the norm may not
imply a break in the series), but undeniably reflects a
very real attitude. lidai-geki, gendai-geki and also ken-
geki (sword plays, a more pedantic term for chambera)
were simply variants on a single object, the film:
Tanaka's The Red Bat was jidai-geki, Mizoguchi's Sisters
of Gion, gendai-geki, and his Tale of Late Chrysan-
themums a Meiji-mono dealing as it does with the Meiji
period that comes 'between' jidai and gendai. Indeed the
Japanese divide their films as rigorously as they divide
the periods of their history or the many natural seasons
of the year. Historical criteria are not the only ones that
are used, of course, but the jidai-Meiji-gendai series may
be regarded as primary to all the others. Like all the
others, these categories are purely descriptive, not
hierarchical; and they are far more widely used by both
critics and general public than any hierarchical distinc-
tions (such as B-picture and main feature, or art-film and
pot-boiler) which dominate our Western discourse on film.
IRON TREES, The other major type of genre classification is based
GOLDEN FLOWERS upon what we might call 'subject-matter'. However, in
this case the term covers much more than the general
plot outline and setting. Seinen-eiga or haha-mono are
not simply films about youth or about a mother, they are
films set in quite specific social milieux, developing a
very similar ideological message, with more or less the
same set of character types and more or less the same
narrative structure. To take a more recent example, prac-
tically all of the yakusa-eiga ('bad-guy movies') made in
the late 1960s and early 1970s - and this means over a
hundred films a year - resemble each other closely in
setting, subject-matter, structure, pace and tone; they pre-
sent, in particular, the same actors in the same typed
roles with only a change of name. 1 Them chambera of the
1920s and the shomin-geki (films about 'townspeople',
i.e. the lower middle-classes), which were the dominant
genres of their respective periods, also share this extra-
ordinary homogeneity, which of course can also be traced
back to the poetry of the Heian period and indeed to
almost every Japanese artistic practice. It is often said
that Bresson, for example, makes the same film over and
over again under different guises; this is the hard core of
the auteur theory. From a Western point of view, this
may be true, but any comparison with the shomin-geki of
Ozu or Gosho makes his work seem a veritable zigzag of
eclecticism. And despite the important ideological and
above all formal differences between the films of these
two men, and between their films and those of Shimizu,
Shimazu and countless others, they constitute a remark-
ably unified genre. They maintain a conformity incon-
ceivable in, for example, the gangster movies of Holly-
wood, with their self-conscious and unrelenting search
1. Of course, many of these films are
for a 'new twist'. As stated earlier, the essence of Mizo-
also part of a series, which further guchi's or Ozu's films is not originality, but a supremely
heightens the sense of sameness.
refined contribution to a continuum.
2. After the Pacific War, the genre may There appeared towards the end of the 1920s, how-
be said to have revived in the films of
such directors as Imai Tadashi and
ever, an important genre which to an extent seems to
Yamamoto Satsuo, though the have escaped from this homogeneity, no doubt because it
development of class struggle was such
that these films reflected far more often
was the first to reflect class struggle rather than the
than in the 1920s the aspirations of the 'unanimist' ideology of traditional Japan. It lasted about
proletariat.
five years, and then succumbed to the repression that hit
3. At about the same time, the the whole radical movement. 2 This new genre, called
Communist Party was able to create a
special committee called Pro kino, for
keiko-eiga ('tendency film') reflected the impact upon the
the distribution and production of films liberal petty bourgeoisie of such pregnant ideas from the
for workers. There were some forty
productions. Of the five surviving films,
West as 'democracy', 'socialism' and even Marxism. 3
I have seen only May Day (I 925) by Gendai-geki, to which the keiko-eiga of course
Iwasaki Akira, a film of newsreel
format, predominantly consisting of
belonged, had developed through the 1920s as an effort
long shots of demonstrators and police. to master and to naturalize the codes of the Western film.
Insofar as they bore strongly the stamp of traditional SOME REMARKS ON THE
attitudes the results remained out-of-phase, de-centred GENRE SYNDROME
with respect to those codes. We are about to examine the
consequences of this. However, when it came to emulat-
ing the social cinema of Europe in films inspired, on one
level or another, by class struggle in Japan, a fresh need
was created for better assimilation of codes so closely
associated with realism in the West. And, of course,
diversification is one of the Western models. Iwasaki
Akira, the celebrated left-wing critic, has described the
two main sub-genres into which these films seem to have
been divided: 4 some 'took the alternative of depicting in
a very real way the miserable petit bourgeois outlook of
the unemployed and wage-slaves euphemistically called
[the] salary-man. Others chose to handle serious prob-
lems into satirical comedies. The former [i.e. the latter]
group is represented by Ozu Yasujiro who directed The
Chorus of Tokyo (1931) and I Was Born, But . .. (1932);
the latter [former] by Uchida Tomu, the director of The
Champion of Revenge (1931), and Itami Mansaku who
directed The Unrivalled Hero (1932).' 5
Besides this general subdivision, it seems that at least
the 'serious' tendency films were more differentiated
individually as well, if only in accordance with the
stereotypes of Hollywood and Europe. The half-dozen
films of this sort which I have seen including Ozu's That
Night's Wife and Mizoguchi's Tokyo March seem to con-
firm this hypothesis.
Richie and Anderson explain that the tendency film
reached its peak in 1932 and then, due to frequent cen- 4· See Japan Film Yearbook, 1937.1! is
sorship, degenerated politically into bedroom farces significant of more than one
contradiction in the Japan of those years
about the lower classes. Others took the form of that nearly all of the articles in that very
rumpen-mono, films about the lumpen proletariat, a less official publication have, overtly or
between the lines, a Left bias. Such a
dangerous topic than bona fide and possibly organized situation would have been
proletarians, less subversive than holding a mirror to the inconceivable in Nazi Germany or
Fascist Italy.
exploited lower middle classes. This genre produced,
however, important works of art, including Ozu's An Inn 5. These films of Uchida and ltami
(Adauchi senshu and Kokushi musii)
in Tokyo (1934). have not survived, but Uchida's Police
The entire history of Japanese cinema, like the whole (Keisatsukan, 1933) is a perfect
pastiche, well ahead of its time, of the
of Japanese art and life, is intricately patterned by genres post-war Hollywood police
and sub-genres. 6 We will return to them incidentally, investigation film with social overtones.
even though the main thrust of this study is in the direc- 6. For a fuller listing and description, the
tion of the deeper structures of Japan's cinema. For, as I reader is referred to Richie and
Anderson (particularly Chapter 13). On
have been at pains to insist, Japan, viewed in and for this subject they are quite informative, if
herself, is an assemblage of just such surface patterns. whimsical.

153
 16. Ozu Yasujiro

Ozu began making films in 1927. With the exception of his

very first film, Sword of Penitence (Zange no yaiba, 1927),
he is known to have specialized from the start in gendai-
geki. Only six of the first eighteen films which he had
directed by the end of 1931 survive today in part or in
whole. I have seen only three of these, but Days of Youth
(Wakaki hi, 1929) and The Lady and the Beard (Shukujyo
to hige, I 930) are no doubt good examples of a naturalized
form of the Western bourgeois comedy, subtly off-centre
by virtue of its very slightness, as it was widely practised in
Japan at that time. 1
Whether or not That Night's Wife (Sono yo no tsuma,
1930) resembles other early Ozu, whether or not it resem-
bles other films of the Western 'social' tendency, I have no
way of knowing. The film is, however, fascinating evidence
of the impact which American films and Western culture
had, not only upon Ozu but upon a sizeable portion of the
Japanese middle and lower middle classes. It is an adapta-
tion of a piece of American pulp fiction. An unemployed
office-worker commits a robbery to bring food to his starv-
ing wife and medicine to his sick child. That same night, a
plainclothesman comes to arrest him but the wife manages
to get her hands on his revolver and the trio spend the night
in tense confrontation, with the detective held vigilantly at
bay. Overcome by fatigue, the couple doze off and awake
1. I have not been able to see the
recently discovered Be Cheerful
to discover the detective has the upper hand again. How-
(Hogaraka ni susume, 1930) or I ever, his heart has been softened by his glimpse of the
Flunked, But ... (Rakudai wa shita
keredo, 1930), nor have I seen the
man's personal tragedy and he gives him a chance to
twenty-minute fragment of the earlier I escape. The husband nobly prefers to pay his debt to
Graduated, But ... (Daigaku wa deta
keredo, 1929).
society; he and the detective march off side by side into the
dawn.
2. As familiar to the Japanese as to us.
Unjust suffering, revolt and penitence
In addition to this familiar pattern 2 the film, though
before the social order are an clearly set 'in some Japanese city', scarcely shows us a
established theme in Japan ( cf. the
Chiishingura theme). Ozu's choice of
single traditional artefact (the couple's tenement room has
this obscure piece of hack-work clearly Western beds, chairs, etc.), and the few studio exteriors
repeats the pattern of Japan's discovery
of Western literature according to a
show an anonymous concrete metropolis, close to the set-
principle of recognition. tings of The Crowd or Spione. The film reveals, moreover, a

154
thorough mastery of the Western system, with only an ozu Y ASUJIRO
occasional false eyeline-match (though we shall see the
importance of this apparently insignificant detail). There
are a great many descriptive pan-shots, aimed at stressing
the squalor of the principal set. There are also, however,
close-ups of objects which do not possess the same con-
notative weight, which appear rather to serve as sheer
transition. The most typical example is a series which, near
the beginning of the film, affords a transition between the
sick child in her bed and the father crouching in the shadow
of a building prior to the robbery.
- close-up of the ceiling-lamp above the girl's bed
potted flower on the window-sill
tree foliage and a street lamp
- high-angle shot of shadows of foliage projected on the
balcony with the sidewalk in the background below
Though this procedure is somewhat simplistic in its linear-
ity, it very directly prefigures a strategy that soon became
essential to Ozu's film-making.
There are also a great many 'spoken titles', but the film
paradoxically pays tribute to that significant avatar of
dominant ideology in the West, the doctrine that 'the pic-
ture should tell the story'. For these titles are totally redun-
dant, so explicit is the imagery, so carefully spelled out in
visual terms is the narrative. Not, of course, that directors
such as Murnau, Leni or Lubitsch, despite their frankly
condescending attitude towards a public judged capable of
perceiving only one idea at a time, actually doubled the
silent pantomine with equivalent titles. On the contrary,
the films which best illustrated the ideology of the pictorial
narrative - such as Der Letzte Mann and Menilmontant -
eliminated titles altogether.
In Ozu's film, which otherwise reproduces the codes of
the Western all-picture narrative, titles re-enact, at the
surface of the screen, the pre-emptive role of the benshi
(who was certainly part of the performance). Possibly the
great explicitness of the images and tautological character
of the titles were intended as 'benshi repellents'. Whatever
the case, the narrative saturation of these images situate
this film at an opposite pole from Ozu's mature works of
the mid-1930s.
Of Ozu's silent films, those best known in the West today
are three comedies made during the next three years:
Chorus of Tokyo (Tokyo no gasshO, 1931), I Was Born,
But ... (Umarete wa mita keredo, 1932) and Passing Fancy
(Dekigokoro, 1933). These films were experiments for the
masterpieces to come. In them Ozu appears to be eliminating
gradually a certain laxness in shot composition (particularly

155
IRON TREES, noticeable in That Night's Wife). It is also in these films
GOLDEN FLOWERS that the systematic low-angle frontality, which has elicited
so much comment from critics, begins to appear. More
generally and more significantly, we witness the gradual
abandoning of that narrative density so typical of the West,
that relentless chain of diegetic events codified by Griffith's
generation and brought to the peak of perfection by Lang
in the first Mabuse. To Westerners used to that narrative
density, Ozu's films, like so many Japanese films, will seem
increasingly 'halting', 'aimless', 'vague', full of 'dead spots'.
Abandoning certain Western codes, however, did not con-
stitute for Ozu an end in itself. He does not stop at refusal,
at a formless 'de-construction'; he organizes the resulting
gaps and ambiguities with a refinement attained by none of
his contemporaries. The dilution of narrative is associated
with a growing stylization of editing procedures. Camera
movements are subjected to a swift radicalization through
numerical reduction, specialization and geometrization.
The pan-shot, in particular, was soon to be eliminated
completely. Ozu's comedies of this period, while perhaps
recalling American silent comedy, are significantly closer
to the Soviet school (Mr West, Bed and Sofa). They display
certain attitudes towards editing (an awareness that the cut
was not to be 'taken lightly') and towards camera move-
ments (an urge to use them sparingly) which were common
to some of the most advanced French and German as well
as Soviet directors during the late silent period.
By these Western norms, Chorus ofTokyo- a portrait of
a white-collar worker and his family - is the most patently
recherche of these three comedies; it is also the closest to
slapstick. The celebrated gag of the wad of banknotes that
falls into the gentlemen's urinal and then is spread out all
over the office to dry is not entirely typical of the film,
whose overall structure does, nonetheless, hinge on a suc-
cession of gags, however slight in some instances these may
be. Far more discreet, and at the same time representative
of the editing subtleties which already point to the future, is
the following gag sequence that takes place in the main
character's home as he is dressing to go out.

- close-up of a table-top with a hat and mirror; the man's

hand enters the shot and picks up the mirror
- medium close-up of mirror as he polishes it on his trouser
leg, crossing the room
- r8oa reversal: medium close-up as he stands knotting his
tie, then looks off-screen left
- low-angle shot of factory chimneys against the sky
- medium shot of main character from behind; the ball is
thrown into the shot from the left
- close-up of ball bouncing on tatami ozv YASUJIRO
- repeat of previous shot; he picks up the ball and throws it
out of shot again (left)
- medium shot of child catching ball from off-screen left
('wrong' direction match)
- return to shot of father; child enters shot from left

Now, while the 'one-idea-per-shot principle' is still main-

tained in this brick-on-brick editing scheme, the sequence
is revealing in several ways. First, it attests to a tendency to
dilute narrative consistency, which becomes increasingly
noticeable in Ozu's work. Late silent films like An Inn In
Tokyo and A Story of Floating Weeds will be almost exclu-
sively composed of such 'mood pieces'. Secondly, the shot
of the sky with chimneys, though tenuously 'justified' by a
glance at what retrospectively may be assumed to have
been a window, prefigures, by its composition, placing and
quiet disruptiveness, the kind of de-centering through
more or less unsituated, off-screen elements which will
soon develop a complex semi-autonomy, as we shall see.
Although the incorrect direction match may seem an insig-
nificant lapse, it heightens the autonomy, the 'otherness',
of the space in which the cutaway shot places the child; it is
also related to the function of the exterior cutaway and has
a similarly premonitory significance with regard to future
films.
I Was Born, But . .. and Passing Fancy are both pitched
in a somewhat lower key than Chorus of Tokyo. The plot
lines are even more slender and the gags far less gross. Both
are centred around children. The former shows the tribula-
tions caused their lower middle-class parents by a pair of
particularly obstreperous but delightful brothers. The lat-
ter deals with the relationship between an unemployed
labourer (played by Sakamoto Takeshi, one of the most
remarkable members of Ozu's growing troupe of actors)
and the young son whom he is trying to raise in precarious
circumstances. All of these films show Ozu's preference for
strict formalization which will develop into the unparall-
eled beauty of the films of the later 1930s and early 1940s,
before its gradual petrification in academic rigidity during
the post-war era. Yet he is still, one senses, casting about
for new strategies. The limitation of camera movement, for
example, produces interesting attempts to use tracking not
merely as a descriptive device, but as an organizational
procedure. The barber-shop sequence in Passing Fancy is
bracketed by slow, contrary tracking shots combined with
sophisticated ellipses through dissolves. All such punctuat-
ing 'opticals' will soon disappear completely for reasons to be
discussed later, but this architectonic use of lateral tracking

157
IRON TREES, will recur again in A Story of Floating Weeds, where it
GOLDEN FLOWERS has much the same function, but as part of a sustained
pattern articulating the film as a whole. Similarly, the use of
a repeated gesture associated with a recurring frame (the
main character in Passing Fancy putting on or taking off his
coat) to give rhythmic pattern to a series of brief sequences,
is another type of structure that will reappear in later films,
though much less obtrusively and, again, as part of their
basic structure.
Although Passing Fancy is said to have been made a few
months after Woman (or Women) of Tokyo (Tokyo no
anna, 1933) I find it difficult to believe, on internal evi-
dence, that this was actually the case. True, the rather
sudden change of genre with this film may have accelerated
Ozu's development. Whatever the case, Woman of Tokyo
is the earliest film in which two important components of
Ozu's system of representation are completely in place, and
they will remain constant in his work until the very end of
his career.
Although its theme is thoroughly Japanese, the film con-
stitutes a rather strange return to the type of plot and
characterization of That Night's Wife. A poor student and
his sister share a flat. He does not know that in order to
finance his studies, she supplements her meagre secretarial
earnings by working nights as a taxi-dancer/prostitute. The
young man's fiancee accidentally discovers the truth (her
brother is a police official and the police seem to have been
particularly concerned with private morality at that time)
and unthinkingly blurts it out to him. Unable to bear the
shame, he commits suicide. 3
Two major aspects of Ozu's systemics 4 figure promi-
nently for the first time in this film. Their theoretical impli-
cations are of great importance.
The reverse-field figure with matching eyelines (and an
acute angle of incidence to camera axis) was not merely the
last component of the dominant Western editing system; it
was, as well, the most crucial. It was this procedure which
3. According to Sa to Tadao, author of a
major study on Ozu and his work, the
made it possible to implicate the spectator in the eye con-
'original' script indicated that the girl tacts of the actors (and ultimately in their 'word contacts'),
worked nights . ' .. to pay her
membership dues in the (outlawed)
to include him or her in the mental and 'physical' space of
Communist Party. Ozu preferred to the diegesis. Clearly such a procedure was basic to the
avoid a clash with the censors. While
this might have made the film a good
illusionist fantasy/identification situation. For many
deal more 'original', not to say bizarre, I reasons, as we have seen, the Western representational
cannot see that it would have made it a
greater step towards An Inn in Tokyo or
system as a whole had gained very slow acceptance in
Only Son . .. norfurthered the interests Japan. Reverse-field editing, in particular, was still being
of the working-class movement.
awkwardly and sparsely used by a large majority of Japan-
4· An association of inter-related but ese film-makers when Ozu began his work. He himself
semi-autonomous systems. The concept
is also useful in dealing with such
assimilated the Western codes of decoupage, handling
Western directors as Dreyer or Godard. them with increasing ease as he went along. One anomaly,
however, persists all through the comedy period; he OZU YASUJIRO

seemed never quite 'sure' about or 'properly' concerned

with eyeline and screen-direction matching. Sometimes
these were 'right', sometimes not. Since this was still com-
mon in Japan and not uncommon in Europe and even
America, this might not seem extraordinary. In 1933, how-
ever (in my opinion, and despite the lost films, it was in
Woman of Tokyo) he began systematically to set up his
camera in such a way as to produce invariably incorrect
eyeline matches. Moreover, he continued this practice until
the end of his career. The angles of incidence nearly always
match; the directions never do. Once one is aware of this
trait, one necessarily sees it as the consequence of a delib-
erate choice, as part ofa fundamental textual economy. This
conclusion is borne out, with a disarming directness, by the
following anecdote, reported by Sato Tadao. 5 At some
point during Ozu's later career (after 1938; unfortunately,
the exact date seems unknown), Ozu's editor, Hamamura
Yoshiyasu, decided it was time to call the master's attention
to the fact that his eye line matching was 'wrong'. In order to
settle the argument which naturally ensued, Ozu decided to
try it - both ways. He shot a two-person reverse-field
sequence as he normally did, matching the eyelines 'incor-
rectly', then shot it again the 'right' way. When the two
montage sets were screened, Ozu's only comment is said to
have been, 'There is no difference.' Hamamura could only
let the matter drop, while Ozu went on imperturbably
setting up reverse-fields 'the wrong way'. I will not presume
to interpret Ozu's cryptic 'There's no difference.' He had
already, at a crucial stage in his career, made a definitive
choice between the two procedures, after using them inter-
changeably. The choice must be regarded, therefore, as
part of a coherent economy within which there was a differ-
ence. The fact remains that by this choice Ozu symbolically
(in the strongest sense of the word) challenged the two
basic principles of the dominant Western mode of re-
presentation. He challenged the principle of continuity, for
the 'bad' eyeline match produces a 'jolt' in the editing flow,
a moment of confusion in the spectator's sense of orienta-
tion to diegetic space, requiring a moment's readjustment.
The resulting effect of hiatus emphasizes the disjunctive
nature of the shot-change, which the developed 'editing
rules' had perceptually obliterated.
Even more fundamentally, by undermining the veri-
similitude of face-to-face reverse-field situations, Ozu
challenged the principle of the inclusion of the viewer in the
diegesis as invisible, transparent relay in the communion of
two characters. Once the spectator is unconsciously obliged 5· Unpublished English translation of
his Ozu Yasujiro no gei·jitsu, Asahi
to rectify with each new shot-change his mental position Shinbun, Tokyo, 1971.

159
IRON TREES, with respect to the players, the trap of participation no
GOLDEN FLOWERS longer functions in quite the same way.
Various explications of this development have been
proposed by Japanese critics. Most of them are postulated
upon the idea that it is a device meant to express 'incom-
municability' or some such neo- Western cliche. Sa to Tadao
offers, somewhat guardedly, a slightly more sober version,
suggesting that Ozu's characters speak to themselves rather
than to their partners. No single 'explication' can satisfac-
torily account for Ozu's decision. Its full implications can
be grasped only by situating it within the whole set of
seminal choices which subtend Ozu's mature work. Once
described, they are seen to offer a discreet but devasta-
ting critique of the fundamental tenets of the dominant
cinema.
Apart from the young Naruse, no other Japanese direc-
tor seems to have imitated Ozu in his systematic disregard
for eyeline matching, not even those who, like Shimizu,
were otherwise indifferent to the Western codes. One can-
not therefore simply claim that linear spatial orientation is
not 'natural' to the Japanese as it is to Westerners. It has,
however, been suggested that for the Japanese 'the entire
experience of space in the most essential respects is differ-
ent from that of Western culture.' In the Ryoanji rock
garden in Kyoto, 'the grouping is such that no matter where
one sits to contemplate the scene, one of the rocks that
make up the garden is always hidden'; the Japanese
'believe that memory and imagination should always par-
ticipate in perceptions. ' 6 This implies that any spatial
arrangement is a 'text' that requires a reading; it is not
simply given and received at face value. Ozu's approach, not
only to eyeline but also to screen-direction and -position
(the 180° match was to become increasingly prevalent in his
work) does express a typically Japanese approach to the
perception of three dimensions by stressing that their re-
presentation is not to be taken for granted.
We shall see how this reading tallies with other aspects of
Ozu's work. For the present, we must examine the second
feature of Ozu's systemics, which had reached full
6. Hall, Edward T., The Hidden development in Woman of Tokyo: the very particular use
Dimension, p. 153. of 'cutaway still-lifes'. (At times, landscapes fill the same
7. 'Makurakotoba or pillow-word: a function, and in the silent films, the 'still' -lifes are occasion-
conventional epithet or attribute for a ally filmed by a moving camera, as in the opening shot of
word; it usually occupies a short,
five-syllable line and modifies a word, this film.) The particularity of these shots is that they sus-
usually the first, in the next line. Some pend the diegetic flow, using a considerable range of
pillow-words are unclear in meaning;
those whose meanings are known, strategies and producing a variety of complex relationships.
function rhetorically to raise the tone With some hesitation, I will call these images pillow-shots,
and to some degree also function as
images.' (Brower and Miner, op. cit., proposing a loose analogy with the 'pillow-word' of classi-
p. soS.) cal Japanese poetry. 7 As we shall see, they also play at times
160
the role of 'pivot-words' and for this reason the term I have OZU YASUJIRO
chosen is not entirely adequate. 8
The pillow-shot may also be regarded as an expression of
a fundamentally Japanese trait. Like Ozu's mismatching, it
is not simply a signature, an individual stylistic trait, but a
culturally and complexly determined sign of dissent from
the world-view implicit in the Western mode. This mode, of
course, is profoundly anthropocentric, as demonstrated by
the rules of centering applicable both to composition within
8. A Japanese critic of the 1940s (Naubu
the frame and to the whole camera/diegesis relationship. Keinosuke) has called these shots kiiten
Prolonged or 'unmotivated' absence of human beings from shOto ('curtain shots'), comparing them
with the punctuating function of the
the screen in a fiction feature film 9 functions as a departure stage curtain in the Western theatre.
from the codes. In Antonioni's Eclisse, this departure is This term has not been generally
adopted in Japan and I feel that it is a
received as a sophisticated poetic statement. In a more misleading oversimplification. The
advanced film like Paulino Viota's Contactos, it has a criti- traditional technical term 'cutaway',
though materially correct, is also
cal, aggressive effect. unsatisfactory by itself, as it does not do
Ozu's pillow-shots have a similar de-centering effect justice to the complex specificity of this
figure in Ozu's work.
when the camera focuses for a moment, often a long one,
on some inanimate aspect of Man's environment. 10 People 9· Unless it conveys some such clear,
unambiguous message as 'Sunday' or
are perhaps known to be near, but for the moment they are 'spooky house'.
not visible, and a rooftop, a street-light, laundry drying on a
IO. The half-dozen films made by
line, a lampshade or a tea-kettle is offered as centre of audio-visually 'illiterate' Navajo
attention. It is the tension between the suspension of human Indians under the non-directive
guidance of the semiologist Sol Worth
presence (of the diegesis) and its potential return which and the ethnographer John Adair offer
animates some of Ozu's most thoughtful work, making a striking similitude with the
non-anthropocentric pillow-shot. In
these shots anything but decorative vignettes. It should be nearly all the films, on-going human
emphasized that while Ozu developed the pillow-shot to a action is frequently inter-cut with
'gratuitous' shots of the surroundings (a
degree of unparalleled refinement, it was used by other stone, a door, a landscape). The makers'
film-makers of the period. Naruse, Shimizu and others general lack of concern with matching,
as well as the actual editing context,
introduce these shots into their films. And from the evi- precludes any reading of these shots as
dence of Tasaka Tomotaka's earliest surviving film, Town 'clumsy' continuity-saving cutaways.
We are clearly dealing with a cultural
of Love (Ai no machi, 1928), 11 Ozu did not invent the and ideological attitude much closer to
procedure; the future director of Five Scouts anticipated the Far-Eastern view of the relations
between man and his surroundings than
the master's later uses of the pillow-shot by some three or to our own. (See Worth, Sol, and Adair,
four years. John, Through Navajo Eyes.)
Ozu's approach to the pillow-shot cannot be reduced to a I r. A print is preserved in the
simple, endlessly reiterative assertion of a single truth - Cinematheque Royale de Belgique.
that Man is not the centre of the Universe - however I2. The suspension of the diegetic flow
fundamental it may be. These shots intervene in a certain is not unrelated to that suspension of the
production of meaning which
kind of discourse, and each de-centering effect possesses its characterizes the practices of Zen
own specificity. These shots cause a suspension of the Buddhism. Roland Barthes (op. cit., pp.
97- I o I) has set forth the premises of
diegesis; 12 this statement requires qualification. For, as this materialist reading, while
analysis will demonstrate, while these shots never contri- commenting upon the annexation of
Zen by neo-Christian ideology as an
bute to the progress of the narrative proper, they often 'answer', proposed by a certain
refer to a character or a set, presenting or re-presenting it Anglo-American bourgeoisie, to the
unbearable tensions associated with the
out of narrative context. The space from which these refer- crisis of State Monopoly Capitalism.
ences are made is invariably presented as outside the Paul Schrader's Transcendental Style in
Film is a recent attempt to conscript
diegesis, as a pictorial space on another plane of 'reality' as Ozu's films into this neo-spiritualist
it were, even when the artefacts shown are, as is often the trend.
a; ~:~t ·:k.:;;.: b~H (:J~~- C-3~~]~-"~-~-~ ir,)
r-s: ~r~:--;~- -~:. ;;·,c:-l~L~L.:.• J ;-~tr~.l~n'< ~-\y -~:.. --~t;·~ .t~,_
,.__ qJ.rn ic: si.,;,;tc~·: s~-<~~," Tn1·ns ~~nd r~~ tCh~~:; 1,
there or are there not people 'present' off-screen; if so, ozu Y ASUJIRO
where? how many? who?) is an essential dimension of the
pillow-shot. The third shot in the series contains, moreover,
an element of ambiguity, likely to be read only in retros-
pect. Such effects of trompe-!' oeil are not uncommon in
Ozu's films. As motionless as those that precede, the tab-
leau that appears here also seems 'uninhabited' at first; it,
too, seems to present 'abandoned' artefacts, like those of
the previous shot. Yet the trouser-leg in the background is
actually 'alive'. Or rather, and here is the crux of the
matter, it comes alive in the next shot, so that the previous
shot is perceived retroactively as ambivalent, as an inter-
locking pivot or transition between the stasis of the pillow-
shot and the 'life' of the diegesis.
This use of the procedure is particularly remarkable in
An Inn in Tokyo (Tokyo no yado, 1935), another portrait of
an unemployed worker (again played by Sakamoto
Takeshi), this time with two boys to raise. It is probably the
masterpiece of Ozu's silent period. Very near the beginning
of the film, after a brief sequence in the industrial waste-
land which is the rather abstract main setting of the film,
there occurs a pillow-shot of a lamp-post with a poster on it
advertising 'Rabies-prevention week'. This announcement
plays a role in the future development of the slender narra-
tive but no one is seen looking at the poster, either before
or after its presentation. For this and other reasons
adduced below in discussing the narrative function of cer-
tain pillow-shots, it must be considered, at one level, as
equivalent to the next shot in the series, which shows the
signboard of the inn, occupying a discreetly peripheral
position in a beautiful window composition with a lantern.
The next shot continues with the utter stillness of a pillow-
shot, but while the centre of the screen is occupied by one of
Ozu's 'signatures' (a certain form of glass lampshade ), in
the lower right-hand corner and completely out of focus (it
is the lampshade which is sharp) is the motionless face of a
small boy asleep. We next cut to a closer shot of the same
boy, still asleep and still absolutely motionless. The next
shot shows a man on a bed, also asleep and motiorless, but
in the foreground and in focus, while in the background,
out of focus, are the father and his two sons, equally unmov-
ing. At last, one of the sons almost imperceptibly moves his
arm, but it is not until the sixth shot in the series- the first of
several close-ups which follow- that full movement finally
returns, and the diegesis acquires its usual consistency.
The crucial role of the hard-and-soft focus relationship
within such 'pillow-structures' is further illustrated by the
last three shots of the opening sequence in Woman of
Tokyo described in part above. After her brother has gone,
3a
Fig. II. Ozu Yasujiro, Woman the sister potters about the room (Fig. I I, Pl. I a), removes
of Tokyo her smock (PI. I b) and finally settles down at the far end of
the room before her make-up box (Pl. Ic). She is quite out
of focus, but the camera remains in this position for nine
seconds. We note the looming presence in the foreground
of objects such as tea-kettle and rice-cooker, all typical
pillow-shot objects. 17 They have figured, or will figure, in
17. And all emblematic of the Japanese pillow-shots in this film. Their presence here and the fact
home. Nearly all the pillow-shots carry that the sister can scarcely be seen to move during the final
with them the cluster of incidental
symbolism traditionally attached to nine seconds (her apparent stillness is heightened by the
nearly every object- plant, tool,
decoration, etc. which is part of the
scale and low definition of her image) make this one of
traditional way of life. those 'semi-pillow-shots'. This 'pillow-status' is further

r68
emphasized by the fact that the focus is such as to blur the ozu Y ASUJIRO
human centre of the diegesis (soft-focus in the pillow-shots
is often used for this kind of de-centering).
As pillow-shots come in series, this ambivalent image is a
prelude to the sequence's closing shot. After an intervening
close-up in which movement, definition, etc. ensure 'nor-
mal' diegetic production (Pl. 2), there is a close-up in the
mirror of the sister completing her make-up (Pl. 3a). We
now see how an 'animate', fully diegetic shot can become a
pillow-shot: the sister leaves the shot, the mirror remains
empty (Pl. 3b) and this out-of-focus, reflected image is held
for five seconds, so that the shot is quickly transformed into
an inanimate pillow-shot - an effect facilitated, of course,
by the lack of any off-screen sound. The next shot is sepa-
rated from the previous one neither by dissolve, fade nor
other 'punctuation'. Even at this early stage, Ozu's elimina-
tion of such devices was essential for developing the full
ambivalence of the pillow transitions. The next shot in the
series, then, seems entirely unrelated to the empty mirror:
we are still in the frozen space-time 'outside' the diegesis.
Only in retrospect does this 'still-life with gloves' (Pl. 4)
evoke a narrative 'elsewhere and elsewhen' as we discover
the presence and profession of the man who has just shed
them: a policeman come to see the sister's employer (Pl. 5).

4 5
The evocative role of the pillow-shot is rich and varied. It
may refer to a character, as in this instance, or to a setting,
as with the gas-burner and tea-kettle which serve through-
out as introductory emblems of the brother's and sister's
flat. It may also be used as authorial signature; such are the
glass lampshades which appear in so many Ozu films (two
are to be seen in Fig. I I, Pls. I a-c, although in this film the
object does not appear in a pillow-shot until the very end).
One very sophisticated use of the pillow-shot as recurring
emblem of a family and its house is found in one of his last
masterpieces, Toda Brother and Sister (Todake no kyodai,
1941 ). A motif of potted plants and a caged bird reappears
IRON TREES, in a variety of compositions throughout the film. In this
GOLDEN FLOWERS instance, particularly close to the actual pillow-word of
poetry, the emblematic function is more general, more
ambivalent than that of the gloves in Woman of Tokyo, or
of the sword in its scabbard which introduces the sequences
in the home of the fiancee's policeman brother.
And indeed, from the early to the later films, the rela-
tions of the pillow-shot to the diegesis evolved and became
less specific. It might be objected, for example, that the
shots of drying socks (Fig. IO, Pl. II) and of a tree against
the sky (Pl. 17) are not pillow-shots at all but an integral
part of the diegesis. In terms of visual style, both of these
shots are pillow-shots. However, unlike the 'purest'
pillow-shots, which transmit no diegetic information
beyond the suggestion of a timeless place or presence, these
shots do participate in the diegetic movement: they
respond to what preceded ('yes, there are socks', 'the sister
sees that it is a fine day'). Moments in which 'the picture
tells the story' were already rare enough in Ozu (the only
other one in this sequence being the close-up of the hole in
the sock, Pl. 9). Yet we do find a few more such instances in
the silent films, such as the bicycle in a house entrance,
indicating the son's presence in his foster-mother's house,
in A Story of Floating Weeds, and this later narrative
annexation of the pillow-shot is actually more systematic
than the examples in Woman of Tokyo. The shot of the
socks, in its second appearance (Pl. 15), is diegetically
superfluous, and is thus partly expelled from the diegesis.
The bicycle, on the contrary, while it conveys each time the
same message of presence, brings fresh temporal data to
the progress of the narrative ('this is another visit'). How-
ever, this direct narrative function was gradually phased
out, and disappeared completely in the sound films, which
is not surprising.
Another sequence in Woman of Tokyo seems worth
citing here, precisely because of the way in which semi-
narrative functions are incorporated into a pillow-
structure: at the end of the film, when the sister has come to
the fiancee's flat to find out if she has seen the missing
brother, a child appears to tell the fiancee that she is wanted
on the phone. The fiancee goes out and the sister is left
alone.

- she glances to her left

- Ps: kimono hung up to dry, framed in the doorway to an
adjoining room
- she looks up
- PS: in the right foreground, out of focus, one of those
'signature' lampshades and in the left background, in focus,
a clock indicating 8:20 (The information offered as to the ozu YASUJIRO
time of day is the diegetic centre of the shot- the presumed
object of the sister's gaze- but while this is made explicit by
the focus, it is contradicted in part by the predominance in
the composition of the lampshade)
- Ps: interior of a watch-maker's shop: clocks all over the
wall and, in the foreground, a telephone receiver off the
hook; the fiancee enters the frame, smiles a thank you at
someone off-screen and picks up the receiver
(A titled, cross-cut sequence sets forth a conversation with
her brother in which he tells her of her fiance's suicide)
- numbed, she slowly hangs up the receiver
- PS: composition of clocks on wall
fiancee standing by phone with hand still on receiver
PS: clocks on wall (different shot)
Ps: clock and lamp seen earlier in fiancee's flat (this time
the composition is laterally reversed, and the lamp on the
left is now in focus, while in the background to the right, the
clock, still indicating 8:20, is blurred)
- Ps: (actually a 'false' one, not unlike the second shot of
the film) shows the traditional charcoal brazier with tea-
kettle and, in the foreground, the motionless pattern of the
sister's kimono
- in front close-up, the sister looks up as the fiancee opens
the door

Several observations are in order here. The shot of the

kimono drying and that of the first lamp-with-clock com-
position are both, through the sister's off-screen gaze,
related to the diegesis. Visually, however, they partake of
the timeless stillness of the pillow-shot, and their diegetic
relationship is in no sense part of a chain of necessary
meaning, like the socks-shot analysed earlier. Thus, for a
moment, the diegesis is suspended, only to 'recover' as we
cut to the telephone-with-clocks. The cutaway to the clocks
on the shop-wall as the fiancee hangs up acts as a diversion;
we seem to have entered another transitional suspension,
but the action is then resumed exactly where we had left off.
The shot was in fact a suspensive parenthesis. However, the
appearance at this point of a very similar pillow-shot is, on
the contrary, the first in a transitional series which consti-
tutes a double variation on the earlier series. The cut from
clocks on shop-wall to clock-and-lamp is a signifying
( diegetic) reversal of the earlier transition (it 'takes us
back' to the flat) while the second of these shots is a
non-signifying reversal of the first shot in the earlier transi-
tion (reversal of positions and focus relationship). The
third of these pillow-shots, moreover, is a reference to
another 'pillow-structure' (the tea-kettle and brazier have

171
IRON TREES, been emblematic of the fiancee's room throughout the
GOLDEN FLOWERS film). In this particularly dramatic sequence the suspension
of meaning itself acquires the weight of pathos and,
thereby, a second meaning, a connotation. Such rhetoric
was soon to disappear completely- together with scenarios
like this one. Structurally, however, these early pillow-
shots function according to a model which was to remain
essentially unchanged for thirty years.
The final shots of Tokyo exemplify the pillow-shot at its
purest, as a satori-like 18 suspension of meaning; they pre-
figure its use in the most mature films. Scandal-mongering
reporters try to interview the sister and fiancee over the
brother's corpse; then we see them back in the street, going
about their business; they come upon a piece of 'real news'
posted on a telegraph pole; they go out of the frame,
chatting merrily. After a few seconds there is a cut to a
close-shot of the nondescript base of the telegraph pole,
then to a dolly-shot along the kerb-stone, in the course of
which the image fades out. Not even the ironic tone of the
first ofthese pillow-shots (the news-bill, the 'emptiness' left
by the reporters) can be said to survive in the semantic void
of the last two, which provide a paradigm of de-centering
and meaninglessness, the filmic equivalent of the Zen-
master's stick-blow answers to his disciples' earnest ques-
tions.
I shall cite further significant instances of the pillow-shot
in dealing with Ozu's later films, but first we must examine
the other aspects of his systemics.
Camera movements, severely restricted in number and
specialized in Woman of Tokyo, were gradually to disap-
pear almost altogether. The successive stages of this disap-
pearance are of considerable interest in themselves. I have
already referred to the occasional moving semi-pillow-
shots in A Story of Floating Weeds. In the subsequent An
Inn in Tokyo, there is still a surprisingly large number of
track shots, but these, without exception, are used to
'freeze' apparent screen-movement; they are follow-shots
or reverse follow-shots that accompany the wretched father
as he wanders aimlessly through the industrial suburbs of
Tokyo with his two boys, occasionally making a half-
hearted attempt to find work. The same frame is main-
tained throughout each given shot. Camera movement is
used to reproduce, at moments when a relatively ample
pro-filmic movement is indispensable to the narrative, that
principle of immobility central to the pillow-shots and
18. 'Satori ... which Westerners can which came to dominate Ozu's work as a whole, especially
only translate with vaguely Christian after he began to work with sound. Paradoxically, this
words like enlightenment, revelation,
insight .. .' Barthes, Roland: op. cit., gradual absorption of the diegesis into the static world of
p. 99· the pillow-shots, though essential to Ozu's development,

172
ultimately contributed directly to the frozen academicism ozu YASUJIRO
of the films that followed There Was a Father, made in I 942.
In Only Son, Ozu's first sound film, the first and almost
the only camera movements appear as a sudden, startling
departure from the principle of camera fixity which charac-
terizes the opening section of the film. These are a series of
gently weaving low-angle shots taken from behind the
mudguard of a taxi which, we soon learn, is conveying the
old mother and her son, whom she has come to visit for the
first time in the city. In Toda Brother and Sister, only a
single camera movement appears, in the film's penultimate
shot, a very brief pull-back when the sister is disagreeably
surprised not to find her brother in his room and realizes
that he has not kept a promise of marriage he has made.
The poetic aptness of this effect is completed by a long shot
of the brother running on the beach. In his last masterpiece,
There Was a Father, he eliminated camera movement
altogether.
This development is not to be explained by any single
reason; these directorial positions form a coherent whole.
However, even as early as Woman of Tokyo, camera
movement is used to contribute to the effect of quietude.
And one reason why Ozu's richest period is that of the late
silent and early sound films, is because of their dialectical
structure in this respect whereas the systematic exclusion of
movement in most of the post-war films seems, ultimately,
an impoverishment.
We must now consider the low-angle camera position
which has come to be identified with the films of Ozu. Of
course, many Japanese film-makers often adopt a low level
for their camera, if only because the Japanese, in their
traditional interiors, spend much of their time on the floor.
Unless one wishes a constantly high angle on seated charac-
ters (like Mizoguchi in his late silent and early sound films)
a shortened tripod is indispensable. Ozu did, however, use
this approach more systematically than any of his contem-
poraries. Many 'expressive' explanations have been
offered, of which the most pertinent is Sa to Tadao's sugges-
tion that this angle endows all pro-filmic activity with a
'ceremonial quality'. This peculiarly Japanese reaction is
perhaps difficult for us to appreciate fully, except insofar as
religious and other ceremonies in Japan do take place close
to the floor and involve low bows and prostration. How-
ever, the key to Ozu's specific, radical use of this procedure
lies in more theoretical considerations. In order to under-
stand it, we must examine the films subsequent to Woman
of Tokyo. For while the camera in this film is often very
low, the type of composition produced often involves an
emphasis on deep space relatively untypical of Ozu's work.

173
IRON TREES, This is partly due to the nature of the lodgings- one-room
GOLDEN FLOWERS flats- in which the film was set. It is also quite clear that the
principles of flatness and frontality had not yet been added
to his gradually evolving systemics. From A Story of Float-
ing Weeds onwards, it will become increasingly apparent
(and overwhelmingly clear in Toda Brother and Sister and
There Was a Father) that a set of converging strategies is at
work. Not only is the camera unusually low, but it is sys-
tematically placed at right angles to the back wall and in the
centre of the rectangular 'vault' of the room or set. The
characters are usually filmed head-on and in long shots.
They generally occupy only the lower part of the frame
when they are seated. This, in combination with the low-
angle frontality, is a further flattening factor, since one
perceives that which is behind the characters as above
them. Also, as Sato Tadao points out, in the later films the
converging borders of the tatami, insofar as they are visible
at such a low angle, are carefully obliterated by the placing
of flat cushions.
Now, all of these strategies produce a single effect: the
elimination of depth indices, the flattening of the image, its
reduction to the two-dimensional surface of the screen. We
have already discussed the surface-oriented character of
traditional Japanese art and architecture and the uncon-
scious tendency to ignore the Western strategies of depth
representation in earlier films. It seems obvious - and an
examination of Figs. 12-16 will prove helpful here- that
Ozu uses every technique at his disposal to produce the
filmic image as picture plane.
Stillness, which was becoming increasingly predominant
in his films, is important in this development, since move-
ment (and especially axial movement, which rapidly
became rare) is basic to the illusion of depth. Once we
understand this preoccupation with a picture plane, other
aspects of Ozu's systemics begin to take on importance.
Series of reverse-angle shots in his mature work are seen as
a succession of flat surfaces, side by side rather than face to
face; there is no imaginary space between for the spectator
to be inserted, as it were. There is no encounter between
eyelines, the necessary complements in Western cinema of
the receding and converging parallels of deep space. Simi-
larly- and this is particularly true of A Story of Floating
Weeds and Only Son- the systematic neglect of direction-
matching (of frame exits and entrances) tends to prevent
successive shots of a given interior from flowing into one
another 'naturally'. The mental reconstitution of a three-
dimensional space on the basis of these 'badly joined', flat
images, requires a considerable effort of memory and
imagination, i.e. a reading.

174
 Another strategy of editing related to the overall flatness OZU YASUJIRO
of image is Ozu's fondness for 180° cuts, in particular
without change of shot size. There Was a Father is rich in
matches of this kind, rigorously proscribed by the film-
school primers, which produce the disconcerting effect of a
picture that has been abutted to its photographic reversal-
the shot seems to 'flip over'.
Finally, though it would be absurd to reduce the pillow-
shot to any 'ultimate' function, the pictorial quality which it
almost inevitably displays may be regarded as the epitome
of Ozu's surface imagery.
Though such metaphors have their limits this visual flat-
ness has a diegetic 'equivalent', for incidents soon grow
rare in the Ozu narrative. These portraits of family life, or
lessons in family relationships, become increasingly event-
less. This phenomenon, which touches a whole sector of
Japanese film-making, will be discussed later. As concerns
Ozu, particularly important are the developments after his
belated conversion to sound. The early sound films (from
Only Son to There Was a Father) quite naturally relied
more on dialogue, but while this did facilitate characteriza-
tion (within, however, very narrow culturally and socially
determined limits), it also soon tended to reduce the narra-
tive incidents to spoken events, and the relative autonomy
of the imagery, mostly of people talking, increases greatly.
Here one may observe an unexpected similarity with the
doll theatre. The wildly contorted features and almost
unbearably expressive vocalization of the gidayu bus hi, the
slow succession of the all but frozen tableaux formed by
puppets and manipulators (the puppets often remain
motionless for minutes on end, with only an occasional bob
of the head and flick of the hand to indicate the identity of
the 'speaker') suggest indeed a parallel: a division of labour
between presentational image and representational voice
in both the performance of the dolls and the films of Ozu. 19
Only Son (Hitori musuko, 1936), the master's first sound
film, is not only his most radical experiment with this divi-
sion between the verbal and the spatia-temporal parame-
ters; it is also, in my opinion, his supreme achievement. It is
difficult to describe with written words and still photo-
graphs this principle of disjunction, as it is most commonly
and reiteratively presented in long dialogue scenes. In the
analysis that follows, this disjunction may be taken as the
fundamental basis on which Ozu has built a system of 19. I have already mentioned one film
'pillow-structures' and other relationships. by Antonioni in connection with the
de-centering function of the pillow-shot,
The sublimated plot of Only Son is a far cry from the and it is not without interest that his first
melodrama of That Night's Wife or Woman of Tokyo. It -and finest-film, Cronacadi un amore,
is based on a comparable dissociation of
tells of a widowed working-class mother who, at the insis- text and images (see my Theory of Film
tence of her son's schoolteacher, makes every sacrifice to Practice, Ch. 5).

175
IRON TREES, send the boy to high school. Years later, when she visits him
GOLDEN FLOWERS in Tokyo, she finds that he is one of the vast city's army of
unemployed, barely able to keep alive on part-time night-
school teaching. Realizing that his mother's disappoint-
ment is justified and a source of profound shame, he prom-
ises to apply himself to getting ahead in the world. His
mother goes back to her province and the silk-mill where
she has toiled all her life; nothing remains for her but to die.
One of the pillow-shot emblems used frequently in this
film is the traditional head-rest or mounted bolster which,
in accordance with polite custom, the son has bought for his
mother's use during her stay in his house. Its use in one
brief sequence, mid-way through the film, is worth describ-
ing here. A pillow-shot of the sleeping mother's kimono
hung on the wall, attesting as part of its function to her
general presence in the house, is followed by a shot in which
the bolster occupies most of the foreground (in focus). In
the background, completely out of focus, the son and his
wife discuss the problems (in particular, the financial ones)
caused by the mother's visit, happy though they are to have
her. The conversation is pursued in a pair of medium
close-ups of the couple (with the 'incorrect' eyeline match
which by now was standard), followed by a slightly larger
shot taken from the same angle, showing as before the
bolster with husband and wife in background. This time it is
they who are sharp while the bolster is out of focus. In this
shot, their conversation is ultimately interrupted by the
arrival (off-screen voice) of the mother.
This remarkably elegant visual structure is derived so
directly from the 'pillow' concept that it seems almost to
develop in a different space from the dialogue, even though
the couple are actually on the screen. In contrast with the
soft focus in the relatively deep sets of Woman of Tokyo,
there is no effect of depth at all. So close are characters and
bolster that the effect is more like that obtained in the
French silent films of the 1920s by artfully smearing vas-
eline on a sheet of glass in order to blur certain areas of the
picture. Here, too, the fuzzy patches do not appear as
'incidental' background or foreground, but as soft areas
sharing the surface of a composition with the hard areas (a
trait to be observed in many pillow-shots, beginning with
Woman of Tokyo).
It is in Only Son that the organization of the pillow-shot
reaches its ultimate degree of sophistication. The opening
sequence of the film, which I have illustrated photographi-
cally, is a remarkable case in point, involving an interesting
variant on the hard-and-soft-focus motif:

- PS: 'still-life with oil-lamp' (Fig. 12, Pl. r)

(a factory near the son's house in Tokyo provides a subtle ozu YASUJIRO
variant). This sound grows louder during the three shots in
the mill, is still heard very faintly during the long shot of the
village; it gives way to another long silence during the last
lantern shot. The series ends with the synch sound of the
grinding stone over which the film's first dialogue will soon
be laid. In this sequence a new dialectic of sound/silence is
added to the visual movement/fixity relationship which
already informed the 'pillow-structure'.
It is used remarkably well towards the end of the
mother's stay in Tokyo, immediately after the film's one
'dramatic' moment- a nocturnal conversation in which the
mother pours out her disappointment to the mortified son
while his wife sobs bitterly on the other side of a shoji.
Following a close-up of the couple's sleeping baby (wife's
sobs continue off-screen), we cut to a pillow-shot which has
been a constant motif: a rather wide view of one corner of
the sitting-room. On the wall, placed upside down as tradi-
tion prescribes, is a good-luck scroll given to the son by his
old teacher, and beneath it the usual charcoal brazier and
tea-kettle, along with the baby's empty bottle. The wife's
sobbing continues for a while longer, then silence. After
some thirty seconds (the shot lasts fifty and is the longest
pillow-shot I have seen) there is a barely perceptible
change in the lighting, followed by the emergence of a
distant mechanical throb. Then there is a pillow-shot of the
mother's bolster and quilt, another of silk skeins drying on
a rack and finally (as the background sound grows slightly
louder) a shot of the mother herself holding her grandchild
in her arms and chatting gaily to it as she ambles back and
forth in a vacant lot.
This extraordinary sequence is one of Ozu's most radical.
The implicit meditation on film-time and the ellipsis, the
interweaving of so many different motifs essential to the
film's diegesis (the sequence is a veritable nerve-centre in
this respect) are typical of this, his most mature period.
Ozu made Only Son when he was thirty-three years old.
Apart from the obvious pressures of the political situation,
the master's two long periods of 'silence'- 1936-42, broken
only by the totally atypical and negligible What Did the
Lady Forget? (Shokujyo wa nani o wasureta ka, 1936), and
1942-8 - clearly mark basic divisions in his development:
growth/plateau/decline.
Toda Brother and Sister (Todake no kyodai, 1941)
together with There Was a Father form indeed the
plateau/peak of Ozu's mature development. Both are
characterized, in particular, by an unparalleled predomi-
nance of long-shots. These are often de-centered, as in
Mizoguchi, when the characters are seated. The closer

179
 t /~'

, ::•'-:'-: ~, :d~v-h-~.::-~~~ ~·:·,:_ l ~';::ten r~l/'

·:1 . . :~rc~ ::;;:.~.-; Lsp:;:,.-·::( ~;.~-upi.·
a:;~ ..:.!r !~~- :_;-~~;~l;, ~~ b::.·'~'.O:: ~--~-!Jf-h~ u~

;·:1",: ·: (,';;f-.· ,\:'ri. -~t< fr;n e;

:l i' ·~~·::.'.c·-.;!--T
I:::~ ~: fL·'-,t•
r.'l
academicism to which the master succumbed soon after the ozu YASUJIRO
war. His refined system was imperfectible by then, and
departure from it was certainly not encouraged by the
alternatives which he saw around him.
Ozu is known to have rehearsed his actors for hours in
order to work every bit of 'spontaneity' out of their acting;
in this, his approach is related to Bresson's. Sa to Tadao also
points out that his characters constantly behave as though
they were being observed by others, that their behaviour is
at all times 'public', we might say 'presentational', and that
Ozu's 'psychology' is systematically superficial. I would add
that Ozu's films, whatever their subject, whatever their
formal organization, always evoke the traditional tea
ceremony and what the celebrated tea-master Okakura
called its 'moral geometry':
Be sure you know
That the tea ceremony, in essence,
Is nothing
But to boil water,
Make tea and drink it. 22

This didactic poem by the sixteenth-century tea-master

Rikyii, is one more reminder of the essentially earthly
character of all Zen-related disciplines, epitomized by
cha-no-yu, the Japanese name for what we call the tea
ceremony (it means simply 'hot water for tea'). The
emphasis, in Ozu's work, on the quotidian, the here-and-
now, on everyday gesture and everyday objects (among
them, precisely, the tea-kettle and charcoal brazier) and on
their fine appreciation- by both characters and 'camera'- is
clearly related to cha-no-yu. Many famous accounts of such
'ceremonies' read like an Ozu script, replete with pillow-
shots and seamless ellipses. Here is a passage from a
description of a private 'ceremony' offered by Rikyii to a
fellow tea-master, Sokyii.

The host and the guest had exchanged a few words, when they
heard the sound of the kitchen door open. As Rikyii explained, a
servant he had sent to Samegai for its noted water had just come
back. Saying he would now get that fresh water, Rikyii lifted up
the kettle and went into the kitchen. Sokyii drew near the fire and
looked at it, now that the kettle was gone. The charcoal, so
arranged as to fit the time of the dawn, was beautiful beyond
words. Looking up, Sokyii noticed reserve charcoal neatly kept in
a container on the shelf. He brought some down, added it to the
fire, and cleaned the fireside with a feather brush. When the aged
host reappeared from the kitchen with the kettle in his hand,
Sokyii explained that he had added more charcoal to the fire as he 22. Quoted in Ueda, Makoto, op. cit.,
thought it would be needed for boiling the fresh water, even p. 96.
IRON TREES, though he was sorry to impair the beautiful arrangement the host
GOLDEN FLOWERS had originally made. Rikyil was deeply moved and complimented
Sokyil, saying that he was delighted to entertain such a thoughtful
guest. As the winter night was long, there was still a bit of time
until daylight. Rikyi.i, saying that he would like to serve tea while
the fire lasted, brought out a light meal. As they sat eating, the
morning bloomed out. 23

However, Ozu is not to be read in the lig~t of a single

historical aspect of the Japanese 'text', Zen/haiku/cha-no-
yu. As we have seen, essential traits of Japanese painting
and narrative arts, and of typically Japanese social
behaviour, are always present in his films. The Japanese, of
course, are fully aware of this quintessential quality of his
work, and for many years steadfastly refused to make any
effort to export his films, or those of his rna jor contemporaries.
It was assumed that these films were 'too Japanese' to be
understood by Westerners. I reserve discussion of the elitist
and equivocal admiration his works have aroused in the
West for a later occasion.
What is one to make of Ozu's known admiration for
American films? However much they may have affected his
earliest work, absolutely no such influence is detectable by
the mid- 1930s, and one is inclined to feel that his relation-
ship with the films of Hollywood and Europe (he particu-
larly admired Ophuls and Lubitsch) was purely subjective.
However, two of his key films, Woman ofTokyo and Only
Son, contain very strikingly used citations from Western
films. In Woman of Tokyo, the brother and fiancee watch a
fragment of Lubitsch's sketch in If I had a Million, while in
Only Son the young man takes his mother to see a Vien-
nese operetta. In the first of these interpolated sequences,
Charles Laughton at his desk is given a message that the
boss wants to see him. He climbs long flights of stairs and
passes through endless offices on the way to the inner
sanctum. This sequence is used by Ozu as a subtly
metaphorical transition (ironical in its attitude to the
Japanese sense of hierarchy) between two pillow-shots,
one of the sister's humble typewriter, which closes the
sequence of the police investigation in the office where she
works, and one of the fiancee's brother's sword and scab-
bard, symbol of his social authority. In Only Son, mother
and son watch a scene from a germanic operetta: a coy
plump 'peasant girl' is pursued through a field of wheat by a
dashing, amorous cavalry officer to the accompaniment of
syrupy music and spectacular tracking. This sequence is
offered as the antithesis of Ozu's manner, and yet the
sequence ends with a close-up of a fallen locket which Ozu
23. ibid., p. g8. contrives to integrate into his editing as if it were a pillow-
shot, for it is immediately followed by the shot of the OZU YASUJIRO

mother's kimono, opening the sequence built around the

bolster, as previously described. In this sequence, Ozu
seems more aware than is commonly admitted that his
departure from the language of That Night's Wife was in
fact a departure from Western approaches. It is as though
he introduces a reference by which to measure the distance
covered, while recalling the cosmopolitan enthusiasms of
his youth.
Rediscovery of the three or four lost films made between
1933 and 1937 is not likely to add greatly to our understand-
ing of the work of Ozu Yasujiro. Judging from the incom-
plete version now available A Mother Ought to be Loved
(Haha o kowazu ya, 1934) is a moderately interesting com-
promise between family melodrama (Woman of Tokyo)
and the more subdued profiles of the later work, with
handsome uses of the pillow-shot. It also has some curi-
ously Sternbergian sequences in a water-front bar, in which
the elder son tries to pry his younger brother away from the
clutches of a prostitute. These scenes are probably the last
traces of any actual Hollywood 'influence' on Ozu.
Barring the unlikely emergence of a lost masterpiece, the
mature work of Ozu may be said to consist, then, of five
films: An Inn in Tokyo, A Story of Floating Weeds, Only
Son, Toda Brother and Sister and There Was a Father. This
body of work is incommensurable with that of any other
Japanese film-maker except perhaps Kurosawa. (Too
much of Mizoguchi is still missing for a meaningful com-
parison.) As a contribution to Japanese culture, however, it
is comparable only to that of the great poets, painters or
sculptors of the past. For like theirs, Ozu's oeuvre is not
merely an individual achievement but, more significantly,
that of an historical and national collectivity.

r8s
17. Naruse Mikio and
Yamanaka Sadao

Japanese cinema in the 1930s is predominantly character-

ized by latent and overt conflict between the dominant
representational modes of the West and emerging Japan-
ese modes. The uneven career of Naruse Mikio (whose
work has, in large part, survived) embodies the intensify-
ing contradictions between Japanese and Western cul-
ture, contradictions which were not confined to the
cinema, but which were to have decisive consequences in
cinema after the 1945 defeat.
Japan's general historical development at this time was
overwhelmingly determined by the rise of the militarist
clique and the approach of the war they were soon to
wage against the Western Powers. Yamanaka Sadao rep-
resents, through his political commitment, his premature
death on the Chinese front to which he had been
assigned in punishment for his political activities, and
through the tragic loss of all but two of his twelve films,
the heroic martyrdom of a tiny minority of politically
conscious democrats.
As we shall see, both men also contributed substan-
tially to what is probably the only lasting achievement of
that bleak era in Japan.

Naruse Mikio began directing in 1930, and his earliest

films known to me, both of them silent, date from 1933:
After Our Separation (Kimi to wakarete) and Nightly
Dreams (Yogoto no yume). Both are essentially what we
know as melodrama, although the melodrama never
seems to have acquired classified status as such. Japanese
genre classifications, as we have seen, are based on class
distinctions, historical period or even narrative structure,
rather than on hierarchical distinctions of content and
tone such as have come to be implicit in melodrama,
slapstick, etc. Most ken-geki had strong melodramatic
themes as we understand them. So did many rumpen-
mono (a genre to which Nightly Dreams can be
attached).

186
 After Our Separation tells of an unhappy love-affair NARUSE MIKIO AND
between a geisha's son and a younger geisha, culminating YAMANAKA SADAO
in her attempted suicide and their separation. Nightly
Dreams is a portrait of a young woman who works as
hostess in a waterfront bar to support her indolent, per-
petually unemployed husband and their child. The hus-
band's sense of guilt towards his wife ultimately drives
him to attempt a robbery. He then commits suicide by
drowning in the harbour, near the bar where his wife is
immersed in the forced gaiety which constitutes her
'nightly dream'. In both these films, Naruse casts about
somewhat desperately to achieve a personal style. The
manner in which he uses his double heritage, Western
and Japanese, is interesting, however. It was, of course,
the heritage of his generation. He had thoroughly assimi-
lated the Western system but unlike Gosho, for example,
he was not content to apply itliterally. In Nightly Dreams
he carries the use of the dramatic track-in to such fre-
quent extremes that one sometimes almost feels a parody
is intended. At the same time, he uses, in both films, a
device closely akin to Ozu's pillow-shot, although three
times out of four, his cutaways are produced by the wan-
dering gazes of his protagonists and are given more
explicitly narrative functions than is ever the case in
Ozu's work. In the same film, when we first see the
young woman returning home, she embraces her child,
then glances at the clock.

- close-up of motionless pendulum with clock-face half

out of frame
- close-up of faded flowers
- close-up of calendar indicating the 12th of the month;
the woman's hand enters the frame and changes the date
to the 23rd

Similarly, when the husband comes home after still

another unsuccessful job-hunt, he sees:

kettle on brazier
pot on stove
child's toy car on table

Naruse's approach to the Western codes could be a sim-

ple magnification of cliche, as description of the rest of
this scene can show. The father winds up the toy car and
it falls off the table. At that very moment, his child is
being struck down by a real car, a scene which is
recounted a few shots later in flashback by neighbours'
children.
IRON TREES, On the other hand, there are several much more ambi-
GOLDEN FLOWERS valent cutaways which act as ellipses within conversation
sequences; these anticipate later Ozu as well as Naruse's
own masterpiece. A charming sequence in a vacant lot,
with the father playing baseball in his stockinged feet
while his child uses a caramel to 'mend' a hole in one of
the abandoned shoes, is somewhat more baroque than
Ozu (a very decorative use is made of unburied cement
conduits); it nonetheless anticipates An Inn in Tokyo to
some extent. 1 The film's aping of Western codes is not
always as inept as I have implied. The bar scenes, in par-
ticular, are pleasantly reminiscent of some of the more
effective films of the late 1920s and are quite successful
in the genre. One suspects that Sternberg's The Docks of
New York made a strong impression on more than one
Japanese director.
After Our Separation contains typical excesses due to a
fascination for the most explicitly pseudo-linguistic aspects
of the Western codes. Among them is a feverish suicide
attempt cross-cut with dancing at a party next door. How-
ever, the film also contains the first stage in the develop-
ment of a system of matching on foreground movement
which becomes a principal strategy in Naruse's first sound
film, his single masterpiece, Wife, Be Like a Rose (Tsuma
yo bara no yo ni, 1935). 2

- a city street seen through a window (frame in fore-

ground): a blind descends and blacks out the picture
- view of rush-hour sidewalk at street level; the crowd is
moving towards and past the camera
- tall office buildings, again seen through a window, but
I. Both films also tllustrate what were
here the frame is not visible; a blind descends and blacks
clearly the limits which film-makers out the picture
faced in dealing with such matters as
unemployment, presentnted as a
- again the camera is at sidewalk level: now the crowd
conjugation of social fatality and moves past the camera and away from it
individual resignation.
a crowded railway platform, the train arrives
2. The lost The Girl They're Talking - a girl stands waiting on a street corner
About (Uwasa no musume) which dates
from the same year (described by Richie
- high-angle shot of a train going by; a blind descends and
and Anderson, presumably according to blacks out the picture (end of opening musical sequence)
contemporary accounts, as 'something
of a triumph of technique over
- long shot: inside a large office, the office boy is pulling
old-fashioned material') is one of the down one last blind, whistling the theme of the musical
films one would especially like to
unearth.
sequence; he picks up a tea-tray
- a smartly dressed young woman says good night to him
3· Is this why it was one of the very few
films of that period to have been
- in the street, we find her whistling the same tune as she
distributed abroad? Or was it walks along
mistakenly assumed that the theme
would seem close to Westerners?
In this splendid sequence, Naruse plays with the potential
4· The contrast is significant with
respect to Naruse's role in Japanese
ambiguities of 'pillow time and space' (non- or 'weakly'
cinema. diegetic time and space) with consummate virtuosity. It is

188
IRON TREES, In the classical reverse-field set-up, the 'over-the-
GOLDEN FLOWERS shoulder' shot is the chief alternative to the 'one-shot'. In
the usual Western approach to this set-up, the 'live' ele-
ment in the foreground (a shoulder, an arm, the back of a
head or an object held in a character's hand) remains
discreetly motionless. As a passive presence, it anchors the
shot, fully establishing the signs of confrontation (generally
before the move into close-up) but is never allowed to
intrude, never, except in sequences of violence, used to
play a key role in the shot-change.
Naruse's development of Ozu's radical attitude towards
the mechanics of matching is of considerable interest, as
evidence ofNaruse's participation at that time in the collec-
tive work of his generation. It also helps to confer upon this
film a singular stylistic authority. A great many shot-
changes are articulated around relatively brusque and
ample movements of these foreground elements, move-
ments which occur perceptibly just before or just after the
shot-change proper. These are not, strictly speaking, cuts
on movement. Even when a direct match is implied (for
example, a character's kimono sleeve is seen to move in the
fuzzy foreground of an 'over-the-shoulder' shot and again
in the sharp background of the reverse angle), the extreme
disproportion of perceptual breadth 5 is anything but an
expression of continuity. Consequently, most shot-changes
within otherwise 'transparent' exchanges such as reverse-
field conversations are signalled, 6 just before or after the
caesura, by a disproportionate foreground event which
pulls the spectator's eye abruptly back to the surface of the
screen, in the manner of the descending blinds in the open-
ing sequence. This procedure must be regarded as doubly
'anti-illusionist'; it emphasizes the discontinuity of the
shot-change and exposes one of the essential depth-
producing procedures of dominant cinema. For in the last
analysis, such is the goal of the 'over-the-shoulder' shot. It
produces an effect of depth by the contrast between an
unfocused foreground presence, peripheral to the spec-
tator's consciousness, and the distant, sharply focused
centre of the diegesis.
5· A hand moving in close-up may cross The film also contains remarkable uses of the cutaway,
the screen in half a second, whereas the closely related to Ozu's use of the pillow-shot, though
same hand seen from a distance may
traverse only one tenth of the screen in generally more disruptive. At one point, during the running
that time. Hence a radical and debate between the father's mistress and his daughter, who
ultimately irreducible discrepancy in
surface velocities. Irreducible but not is determined now to bring her father home, the mistress
undiminishable: Naruse was on the starts to rise from a seated position, presumably to break
contrary contriving to make this effect
fully visible. off a painful conversation. On her movement, Naruse cuts
to the outside of the house, so that the movement is con-
6. Though we must not forget that
continuity is already challenged by tinued by that of her shadow projected on the paper wall.
eyeline 'mismatching'. This shot is part of a series of cutaways from this long
dialogue to the adjoining garden. NARUSE MIKIO AND
Unlike Ozu, Naruse never abandoned the use of camera YAMANAKA SADAO
movement, but he had discontinued a grossly expressive
use of it, as in Nightly Dreams. It was now used for a
sophisticated organization of pro-filmic movement,
exploiting entrances, exits and empty frame to good advan-
tage, both dramatically and decoratively.
The approach adopted in this film involves not only a
specifically Japanese refusal of certain norms of Western
cinema, but also a budding sense of formalization based, in
the Western manner, on the transgression of norms. Unfor-
tunately, Naruse seems not to have seen it as a point of
departure for future work. One feels today that Wife, Be
Like a Rose might have had a place in his work comparable
to that of Woman of Tokyo in Ozu's career. His next
surviving film, TochUken Kumoemou (1936), a turgid bio-
graphy of a famous Meiji performer who sacrifices every-
thing for his art, scarcely seems to be the work of the same
director. One is irresistibly reminded of the chroniques
filmees which Marcel L'Herbier was reduced to making in
those very same years. Naruse's is a laboriously academic,
slightly over-edited film, with a preponderance of medium
close-ups, perfectly correct eyeline matching, and the
actors have obviously been ordered to remain perfectly still
when providing foreground support for an 'over-the-
shoulder' shot. Tsuruhachi and Tsurujiro (Tsuruhachi
Tsurujiro, 1938), a 'family drama' about a brother and
sister team of traditional concert musicians, is somewhat
less painfully familiar to us, if only because of some very
fine music, used at times in conjunction with sound-effects
in a manner similar to later experiments of Mizoguchi.
There are soqte decorative pillow-shots as well.
Even what is generally regarded, and rightly, I should
say, as the director's most endearingly successful film of
that period, The Whole Family Works (Hataraku ikka,
1939), is, on the level of editing, a thoroughly 'Western'
film, however 'Japanese' it may be in other respects. It
inaugurates the delicate dramatic style of Naruse's post-
war films.
As one reviews the Naruse films of this immediate pre-
war period, one concludes that he had made some kind of
'agonizing reappraisal'. He seems to have had serious per-
sonal problems around that time, no doubt aggravated by
the war; from 1941 to 1951 he was almost completely silent.
The films which he made during the 1950S, and to which I
shall refer briefly later, are wholly responsible for his repu-
tation in the West. They are certainly as fine as anything he
made after Wife, Be Like a Rose; they are, in fact, excellent
narratives in shomin-geki form, on a level with the best of
IRON TREES, Becker or Lattuada. By this time, however, the specifically
GOLDEN FLOWERS Japanese mode of filmic representation and the formal
innovations which had characterized that early masterpiece
seemed to be no more than a forgotten memory.

The very remarkable promise shown by Yamanaka Sadao

in the dozen films which he made in the 1930s also went
unfulfilled, for very different reasons; he was killed on the
Chinese front in 1938, at the age of thirty-one. Yamanaka is
often called the Vi go (or the Rene Clair) of Japanese
cinema. It is difficult to analyse or evaluate his work with
any assurance, since only two of his films have survived. 7
Both of these are happily representative in many ways of
what is most Japanese in so much of the cinema of that
period. At the same time, they have a special, personal tone
and style that corroborate Yamanaka's reputation as an
auteur. 8
The earliest of the two Yamanaka films known today,
The Pot Worth a Million Ryi5 (Hyaku-man ryi5 no tsubo,
1935), is a delightful low-key comedy, a delicate mixture of
pathos and satire probably rare in that period of social
tension. Set in the late Tokugawa period, the plot is a
variation on a universally familiar theme. A large, very
plain earthenware jar is valued, for unfathomable reasons,
at the astronomical sum of one million ryi5, and a powerful
lord is determined to add it to his rich collection. Unfortu-
nately, one of his vassals, believing it to be worthless, has
offered it, in a gesture of contempt, to his brother as a
wedding present. Genzaburo, an easy-going, quite unmar-
tial samurai, immediately sells it at his bride's behest to
some rag-pickers, who give it in turn to a little boy, Yasu,
the son of a rice-vendor. The rice-vendor is murdered by
ruffians, and the child is taken in by Okami, who runs an
archery range and tea-room, and her lover, Tangekazen, a
one-eyed ronin, a traditional figure in chambera (Fig. 18).
Having discovered the value ofthe jar, Genzaburo's shrew-
ish wife sends him out on a full-scale search for it, and he
7· According to Donald Richie, in
begins to spend most of each day in the archery range,
private conversation, a third film was making eyes at Okami's lovely assistant. The plot is thick-
located in a private collection a few ened by the ceaseless attempts of Genzaburo's brother to
years ago. The films were dispersed on
the death of the collector and the find the jar, which is, of course, right under the noses of
Yamanaka has yet to be found. everyone concerned. (Yasu keeps his goldfish in it.) Gen-
8. In contrast, I am afraid, with ltami zaburo and his new friends finally discover the truth, but
Mansaku, the other best-known radical agree to keep quiet, giving Genzaburo a pretext to stay
director of the 1930s. Akanishi Kakita
( 1936), his one surviving film as a away from home 'looking for the jar', in other words con-
director ( he scripted films for Inagaki tinuing to enjoy himself at the archery range.
and others), may be an effort to renew
the content of the jidai-geki; it is in no We are clearly a far cry from the austere narrative of the
sense a contribution to the visual and shomin-geki, or the grandiloquence of the average jidai-
narrative style of the period, as were the
films of Yamanaka. geki. We are, in fact, startlingly close to Rene Clair, or even

192
IRON TREES, the child's itinerary are explored thoroughly, one after
GOLDEN FLOWERS another, by Okami and Tangekazen as they look for him.
The European silent cinema, particularly that of Ger-
many, did on occasion create dramatic tension through the
use of the empty frame, as in Lang's Mabuse and Dupont's
Variety. The procedure had always been, and is still, more
frequent in Japanese films. This willingness to stress the
existence of a frame, to leave the shot uncentered (or with
an imaginary, 'off-screen centre'), is related to aspects of
traditional painting, as illustrated by an anecdote recorded
by H. P. Bowie.
The artist Buncho being requested to paint a crow flying across a
fusuma or four sliding door-like panels, after much reflection
painted the bird in the act of disappearing from the last of these
subdivisions, the space of the other three suggesting the rapid
flight which the crow had already accomplished, and the law of
proportion (ichi) or orderly arrangement thus observed was uni-
versally applauded. 9

The Western system abhors a vacuum; the frame empty of

people, a quite ordinary event in the Primitive cinema, with
no special expressive value, was rapidly codified. It came to
signify various forms of suspense and was particularly valu-
able in horror films. More rarely, off-screen action was
productive of a style of understatement (e.g. Nicholas
Ray's They Live By Night, Bresson's Les Anges du peche).
As we have already seen, these were not the attitudes
that dominate the use of the off-screen/ on-screen relation-
ship in most of the films of that period in Japan. It was a
non-committal, 'arbitrary' de-centering that was practised
by Ozu, meaningless except in so far as it emphasized the
materiality of the frame, the picture as design and the
diegesis as a circumscribed space. This spirit is present in
the 'empty frame' strategies of most of his contemporaries.
At one point in The Pot Worth a Million Ryo, Shuchibei,
the rice-vendor, has been attacked and mortally wounded
by ruffians. Tangekazen has found him and brings him back
to the archery range. Inside, the women are clearing away
for the night. There is a knock at the door. Cut to the street
outside; Tangekazen stands at the door, with the wounded
rice vendor on his back; the women let him in and close the
door; the camera continues to 'gaze' at the closed door for
eight seconds, as we hear faint voices, then silence from
inside. There follows another long shot of an empty street,
before the cut back to Shuchibei's pathetic death-scene,
shown in full. In other words, it is not possible to detect any
ellipsis, any significant time-lapse inside while the camera
9· Bowie, Henry P .. On the Laws of
was outside. The empty shot has no coded, rhetorical func-
Japanese Painting, p. 48. tion whatsoever. Simply, Yamanaka shows the bird and the

194
empty space behind it. Like most Japanese artists, he saw NARUSE MIKIO AND
representational space/time as an association of fullness YAMANAKA SADAO
and void, of event and non-event.
Yamanaka's use of 'cutaway transitions' in this film is
frequent and systematic, but on the whole more simplistic
than the example above or than Ozu's pillow-shots. Many
are of the still-life variety, but the image is centred on the
manekineko, the good-luck china cat, which is a continual
object of contention in the archery range, or on other props
peripherally related to the narrative strands. The repeated
shot of omochi, a white, gummy savoury associated with
the New Year season, swelling over the charcoal brazier
during the sequence of Yasu's runaway bid is a somewhat
more developed instance. In every case, the procedure
demonstrates the cultural concern with objects and envi-
ronment also obvious in the long shot and common to so
many films of that period.
Humanity, Paper Balloons (Ninja kamifusen, 1937), the
young director's last work, is far more pessimistic in mood.
Set in approximately the same period in Japanese history, it
is a portrait of a poor section of Edo, with its itinerant
pedlars, shop-keepers and artisans, its petty intrigues and
private tragedies. The slender main plot is concerned with
an impoverished ronin and his mistress, who makes and
sells paper balloons. Desperate for money, he ultimately
becomes involved in a hare-brained kidnapping scheme.
When it fails, his mistress, in despair, kills him in his sleep,
then takes her own life. It should be added that the narra-
tive structure is 'circular', since the film had opened with
news of the suicide of another poor ronin.
Among the most remarkable sequences are those which
take place in the narrow, crowded alleyway in which the
ronin and most of the main characters live (Fig. 19). The
camera angle usually coincides with the axis of the alley,
and the editing takes us up and down it in a series of
concertinas that can hardly be said to match at all; each shot
is at once separate from and identical to the previous shot.
The overall result is quite the contrary of one's expectation
of an effect of depth; it is as though new characters were
continually appearing on the self-same 'screen-prosce-
nium'. In these sequences the decoupage has a rough-hewn
refinement which points ahead, towards the mature style of
Kurosawa.
Concertina articulations are developed to a high art in
the most beautiful sequence in either of Yamanaka's sur-
viving films.
The setting is a street much wider than the alleyway
referred to above. It appears to be a more affluent neigh-
bourhood, though close to the slum where most of the film

195
 h·-
.;··
h 'I ,, ! I I \\' )
,. h
·{!
'
i c ; l ~J I i ~~
' ; l
'
! i
'
 18. On Architecture

Implicit in my account of the Japanese cinema is a recogni-

tion of the importance of the country's traditional architec-
ture. Already fully apparent in Ozu's films, it will appear
again in those of Mizoguchi and Ishida. I therefore turn to a
consideration of some aspects of domestic design.
The traditional Japanese home has been and continues
to be designed in a manner common to all, rich and poor. 1
Its principal characteristics are by now well known to Wes-
terners: semi-mobility of the partitions which separate
rooms, and separate interior from exterior. The internal
organization of space is, within certain limits, quite flexible.
Similarly, interior and surroundings (generally a land-
scaped garden, however narrow) may be related in a var-
iety of ways. The possibilities include complete separation
(wooden shutters drawn over the paper doors (shoji) dur-
ing storm or absence); communication of light only (shut-
ters removed, shoji closed); 'framed' communication (for
both exit and entrance and the creation of 'natural pic-
tures'); and relatively free 'flow of space' between indoors
and outdoors, obtained by drawing back shoji as much as
possible. 2 The range of inter-room communication is
equally varied, especially if we consider the existence of
another kind of sliding door, the fusuma, covered with
1. Modern flats, whether "mansions'
opaque (often decorated) paper. It is also important to note
(two-level constructions) or full-blown
apartment houses, are usually fitted that whatever fixed surfaces there may be- particularly the
with non-connecting windows that do flat false ceilings- they echo the bare ribbing oftheshoji. In
not reach floor-level. They do, however,
conform in most other respects to the fact the concern with apparent structure is so great that
traditional model: tatamis, sliding many elements of house architecture are in fact falsely
partitions, few pieces offurniture, quilts'
spread out for sleeping and stored away structural.
during the day, etc. A surprisingly large Some critical consideration of that notion of the 'flow of
number of Japanese still live in
individual houses (hence the huge space' in Japanese dwellings is necessary at this point.
surface area of a city like Tokyo). Heinrich Engel, one of the foremost Western scholars of
2. The shoji are often replaced today by Japanese architecture, takes as one of the general premises
sliding winC:ows of translucent glass, but for his analysis that 'the eye accepts space architecturally,
many houses, especially in the country,
are equipped with shutters, glass i.e. as being defined in character and purpose, as long as the
windows and shOji, providing an even remaining space-making elements are pointed enough to
greater range of practical and aesthetic
combinations. optically suggest controlled space' and that 'since life in
the dwelling is essentially "immobile" ... rather than ON ARCHITECTURE
"mobile" ... sensitive residential space in its most appro-
priate substantiation is arresting rather than driving; it is
static.' Apropos of the Japanese house, Engel maintains,
contrary to the usual view that its architecture is one of free
spatial flow, that 'the two different orders of space define-
ment and space control- the vertical planes for marking the
individual rooms, the horizontal planes for marking the
house interior- are the keys for understanding Japanese
space. In spite of all openness, the quality of space, though
easily transformable, remains static and crystallinely
defined. There is no interchange of space between exterior
and interior, nor does space flow from room to room; there
is no "continuity of space" ad infinitum in one direction.
Instead there is either a succession of space, or a fusion of
space with two or more units being joined into one space.' 3
Let us now suppose that we have set up a camera in a
traditional Japanese house, at a medium height and equip-
ped with a moderately wide-angle lens. We will discover
that we can close off our field of view to the dimensions of a
medium close-up or open it for a long shot by the simple
expedient of manoeuvring the sliding doors between our
camera and the far end of the house. We may also divide
the pro-filmic field into several rectangular or rhomboidal
cells, of different sizes and at different distances from the
camera. We may open 'windows' into the garden in almost
any part of the field. Characters, when we introduce them
into our set, can be brought into the shot from any side and
at almost any distance from the camera. In theory, at least,
we are dealing with a fully developed deep space, far more
exploitable as such than any but the most cunningly con-
trived Western set. It is this aspect of the space before us
which corresponds to the common opinion that Japanese
space is one of 'free flow'. However, if we look into our
view-finder, we will find that Engel's thesis is also very
relevant. In the composition as it appears, now on the
ground-glass, later on the screen, the overwhelmingly
dominant features are rectangles (or rhomboids, if we are
not at Ozu-like right angles to the set) of all sizes and in all
sorts of patterns, formed by intersecting horizontal and
vertical lines: the frames and ribbing of the shoji, the posts
and lintels, balustrades, lattice-work, etc. Even the fore-
shortened trapezoids of the tatami borders (except in tem-
ples, the mats are staggered in patterns that avoid long
straight lines) are part of the system. Any Japanese film set 3· Engel, Heinrich, The Japanese
House: a Tradition for Modem
in a traditional interior offers Mondrian-like patterns, Architecture pp. 24&-8. I am indebted to
many large or small sections of which may at any time slide one of my students at New York
University, Barry H. Novick, for having
back to reveal a new,framed element: a person, a fragment called my attention to the importance of
of a garden, a paintedjUsuma, etc. And although a shadow this study.

199