計画系 660 号

【カテゴリーⅠ】 日本建築学会計画系論文集 第76巻 第660号，341-351，2011年 2 月

J. Archit. Plann., AIJ, Vol. 76 No. 660, 341-351, Feb., 2011
�

A SEARCH FOR ARCHITECTONIC SPATIAL STRUCTURES

AINSearch for Architectonic
NARRATIVE FILM : Spatial Structures
CASE STUDY in NarrativeARK”
OF“RUSSIAN Film:
From film to architecture : AnCase Study
extended of “Russian
cinematic approach toArk”
architectural space design (Part 1)
From物語映画における建築的な空間構造の分析：
Film to Architecture: An Extended Cinematic Approach to Architectural Space Design (part 1)
「エルミタージュ幻想」によるケーススタディ
映画から建築へ：建築空間デザインへの映画記号論的アプローチ（その１）
�����������������������������������������
������������������������������������

Richard DOUZJIAN ＊ and Teruyuki MONNAI ＊＊

Richard
リチャード DOUZJIAN�Teruyuki
ドゥズジヤン，門 内 MONNAI
輝行 �
������������*,�����**

In this paper, narrative film is defined according to structuralist semio-linguistics, and then its units of significance are laid
out in a preliminary model of elementary units. Then, by applying this definition and segmentation process on the case
study film of “Russian Ark”, it was found that all its significant elements and organizational structures are dependent on,
and function of, the denoted architectonic elements of the actual State Hermitage Museum, St Petersburg, where it was
filmed. This dependency of the filmic structures on architectonic elements allowed the preliminary conclusions that their
similarities with architectonic spatial structures are as follows: 1) architectural spatial units are similar to segments and
syntagmas, 2) their organizational structure is similar to the diegetic spatial structure, and 3) architectonic programmatic
and functional organizations are similar to the narrative organizational structures, or syuzhet.

Keywords: architecture, film, diegesis, narrative, spatial structure, semio-linguistics

��, ��, ����, ��, ����, ��-���

1. Introduction
Narrative film, as an artistic product is to this day and age one of the most powerful means to stimulate imagination, feelings and deep
contemplations in its users (i.e. viewers). As architectural designers are in constant search for inspirational sources in order to produce
newer and more exciting designs, we propose film as such a source. Although some architects have already used and are still using film in
that purpose, their approaches remain more intuitive than pragmatic1.
The objective of this paper is to deepen the understanding of film on all its levels of significance in order to have a better manipulation
of its elements. By doing so, architects would be able to design architectures more loyal in form and significance to their filmic source of
inspiration, and add additional substantial and formal layers to their creations, resulting in objects having multiple readings, becoming
more attractive and appealing to their users.
We also focus on the segmentation and the organizational structures of a case study narrative film, Russian Ark, by using structuralist
semio-linguistics2 that allows us to analyze it from its smallest units of significance to the largest. We use semio-linguistics because it
describes and clarifies how the filmic medium functions and how it proceeds to create meanings. Moreover, the growth of semiotic theory
and the presence of its vocabulary in a variety of intellectual fields confirm the importance of the science of signs, sign systems and
signifying practices as a tool for addressing the semantic riches of extremely diverse cultural forms, while semiotics’ cross-disciplinary
thrust constitutes an antidote to the fragmentation and compartmentalization of intellectual disciplines. Therefore structuralist semio-
linguistics plays a bridging role between film and architecture in the search for their similarities.

2. Elementary Units of Film

2.1. Generalities
In the breakdown of film into its elementary units, we largely base our model on the extensive studies conducted by the film theorist
Christian Metz who is considered the reference in this field. Applying Saussurian semiology to define film language, Metz reached the
＊
Doctoral Course, Graduate Student, Dept. of Architecture and Architectural Engineering, 京都大学大学院工学研究科建築学専攻 博士課程
Graduate School of Engineering, Kyoto
* ����������������� ����University Doctoral Course, Graduate Student, Dept of Architecture and Architectural
＊＊
Prof., Graduate Student, Dept. of Architecture and Architectural Engineering, 京都大学大学院工学研究科建築学専攻 教授・工博
Graduate Graduate
Engineering, School of Engineering, Kyoto University
School of Engineering, Kyoto University, Doctor of Engineering
** ����������������������� Professor, Doctor of Engineering, Graduate Student, Dept of Architecture and
Architectural Engineering, Graduate School of Engineering, Kyoto University

− 341 −
conclusion that most narrative films resemble one another in their principal syntagmatic figures (units which organize spatial and
temporal relations in various combinations); and the true analogy between film and natural language operates not at the level of basic
units, but rather in their common syntagmatic nature3. Metz structured film according to his Grande Syntagmatique model (Fig. 1) that
constitutes a typology of the diverse ways time and space can be ordered through editing within the segments of the narrative film.
We use this model for the following reasons: 1) it is the first most developed
“scientific” attempt to analyze film language, 2) it is still widely used by film
analysts and used as the starting point for many other syntagmatic analyses, 3) it
can be useful even when inapplicable during a film analysis because it reveals to
which degree the given film is far from the classical narrative ones, 4) it provides a
clear example for an analytical methodology of film, and 5) it opens the door for
further development and perfection.
In the addition to the syntagmatic structures of film, we learned of the crucial
roles montage, narrative, diegesis and codes play in the construction process of film,
and so, adding together all these elements we generated the preliminary model in
Fig. 2. This model is not a final one, at this stage it is our own first attempt of
understanding and representing film structures, and it is bound to change and
adapt with the advancement of our research and the case studies we investigate.
2.2. Preliminary Model
We establish that narrative, diegesis, one or more autonomous segments and
codes (specific and non-specific to film) form the highest level of significance of the
narrative film. The narrative is audio-visually materialized by the diegesis (the sum
of the represented denotation4), which in turn is expressed, and cinematographically
structured by means of autonomous segments. Narrative, diegesis, autonomous
segments and their respective units and typologies are articulated together by
means of different corresponding codes (Fig. 2).
The syntagmatic structures are simultaneously different and linked to the general Fig.1�Metz’s Grande Syntagmatique model for
film segmentation taken from Film Language: A
narrative and diegetic structures: they are linked because some of the syntagmatic Semiotics of the Cinema�
configurations that constitute it are only found in relation to the story and more
generally in relation to the actions taking place in the
world of the told story (i.e. in relation to the diegesis);
different because these constructions don’t have the level
of abstraction of general narrative constructions. The
latter constructions are completely independent from the
signifiers5 that convey them, while the syntagmatic ones
are founded on represented (figurative) elements that are
visible on screen.
This means that the correlation between syntagmatic
units, narrative and diegesis is quite fluid: segmentations
can be applied and marked either by taking into
consideration the combination of narrative/diegetic
elements (a deeper structure of narrative film) with types
of punctuating montage units (the optical devices,
elements of a more superficial structure) or by taking into
consideration each of the narrative/ diegetic elements and
optical devices exclusively by themselves.
2.3. Narrative
The general definitions of Narrative can be summarized
as the recounting of two or more events (or a situation or
an event) that are logically connected, occur over time,
and are linked by a consistent subject into a whole. Fig.2�Preliminary Model of Elementary Units of Narrative Film

− 342 −
 To further comprehend Narrative, we adopt the Russian Formalist approach6 because it considers film as a narrative art, and treats
the narrative itself as an object that consists of form and substance. The concept of Form here is a very important motivator, as our goal
is to use film to generate architectural Forms and Spaces; the Formalist approach therefore simplifies the task of narrative analysis on
those levels.
According to Russian Formalists, Narrative is distinguished into two levels: the Fabula and the Syuzhet.
Fabula (or story) is understood as the basic outline or raw material of the story prior to its artistic organization. It is the logical and
chronological cause and effects, and relationships between the characters and/or the events of the story.
Syuzhet (or plot) is summarized as the rearrangement of the basic elements of the fabula. This act can take place by using artistic
narrative techniques such as parallel plots, ellipsis, retardation, reverse chronology, etc. Syuzhet can also be described as the formal
manifestation of the fabula: it can be imagined in terms of form and organization, as it can also be referred to as the structure of the
fabula. Structure in this case refers to the arrangement of and relations between the parts or elements of something complex.
Fabula is therefore the content of the narrative in its absolute non-deformed state, as for the syuzhet, it represents the structure that
re-organizes that content. And this is what interests us in our research and our analyses concerning the notion of narrative.
2.4. Diegesis
The concept of Diegesis7 is very important to film, and is one of the notions that give it its particularities. This concept was introduced
to cinema by Etienne Souriau and later expanded into film theory by Christian Metz who asserts that the diegesis “designates the film’s
represented instance, that is to say, the sum of a film’s denotation: the narration itself, but also the fictional space and time dimensions
implied in and by the narrative, and consequently the characters, the landscapes, the events, and other narrative elements, in so far as
they are considered in their denoted aspect”8; Diegesis is then the audio-visually represented world where the narrative of the film is
taking place in.
Intra-diegetic refers to any and all filmic elements (characters, sounds, music, objects, images, etc.) that exist in and affect the diegetic
world. These elements are considered on-screen when in the visual field of the camera, and off-screen when outside its visual field. On the
other hand, Extra-diegetic refers to any and all filmic elements that exist outside and do not affect the diegetic world; the most common
examples of extra-diegetic elements are the musical soundtracks of almost all commercial narrative films, and the unseen narrator
narrating/explaining events occurring on-screen.
2.5. Segments and Syntagmas
Segments, according to Metz, are delimited, identified and defined based on the following three criteria: 1) unity of action (diegetic
continuity), 2) the type of demarcation (i.e. the visible or invisible optical devices used to separate and articulate the major segments) and
3) syntagmatic structure (the principles of pertinence which identify the syntagmatic type).
Accordingly in narrative films, narrative segments are “blocks” of syntagmas that are joined together containing information relevant
to the understanding of the narrative of the film in total. In most narrative films, these segments are articulated with either optical
devices and/or non-narrative segments (themselves constituted of syntagmas with no narrative significance).
However, the delimitation of segments is not always easy since their limits follow more the filmmaker’s way of using/arranging the
basic film elements than a clear general rule: for example, a fade-in/fade-out is sometimes used to simulate the view from an
opening/shutting eye and not to delimit a segment. Moreover the limits of a segment are also relative to the way every viewer/analyst
reads the film and what he is looking to extract from it.
Syntagma ”is the general term used by Metz to designate the units of narrative autonomy (narrative segments), the pattern according
to which individual shots can be grouped, reserving both sequence and scene to designate specific types of syntagmas”9. Metz’s eight
preliminary types of syntagma are laid out and numbered in Fig. 1.
2.6. Shots and Optical Devices
A Shot (single or autonomous) is a continuous flow of images uninterrupted by editing. A shot doesn’t have any specific duration. It is
only limited and defined by the beginning and ending of the capturing process of the camera (photographic, digital or other). It is also the
largest minimal segment of a film, since at least one shot is required to make a film, or part of it. However, “if the shot is not the smallest
unit of filmic significance, it is at least the smallest unit of the filmic chain” (Metz)10.
The Optical Devices (OD) are series of optical signs used during the editing phase of filmmaking (e.g. fade to black, dissolve, fade-out,
etc.). They are punctuating devices that separate, articulate and mark the beginning and ending of shots, syntagmas and major segments
depending on their contextual use in the film by the filmmaker. And finally, they are elements that are film-specific, independent of any
other system, well defined, easily distinguishable and isolated, punctual, unrelated to any specific film or cultural background, and are
some sort of construction signs.

− 343 −
2.7. Codes
Codes are by definition “constructions that establish a systematic correlation between two systems”11. They are the set of widely
accepted conventions that dictate the use of units/elements of a significant system of film within itself (e.g. intra-narrative) or the
units/elements of a certain system in relation to another one (e.g. diegesis-segments).
Metz argues that like any artistic language, the cinema manifests a plurality of codes, and although these remain somewhat constant
across all or most films, there is no single “master code” shared by all. Metz considers that filmic texts form a structured network
produced by the interweaving of specific cinematic codes and non-specific codes. The specific cinematic codes are codes that appear only in
the cinema, that is to say codes linked to film’s definition as deploying moving multiple images (e.g. codes of camera movement, montage
and continuity editing); as for non-specific codes they are those that are either shared with other arts (e.g. narrative codes, shared with
novels, or codes of visual analogy, shared with painting), or completely non-specific such as codes widely spread in society and culture (e.g.
codes of gender roles).

3. Analysis of Russian Ark

3.1. About Russian Ark
Russian Ark is a 95 minute12 long, single-shot Russian film directed by Alexander Sokurov in 2002. The story consists of the wandering
of two figures inside the St Petersburg’s State Hermitage Museum, Russia: the first figure, the “Stranger”13, is seen by the spectators,
while the other (the camera-figure as he will be further referred to) is unseen, while his voice (that of Sokurov himself), the camera
movements (representing his own visual trajectory in space) and the “Stranger” communicating with him being the only references to his
existence. The wandering of these two figures takes them through more than 40 different spaces of the Hermitage Museum’s Theater and
Winter Palace (where most of our analysis is focused), while interacting with each other and other characters representing 200 years of
selected Russian/Hermitage history (from Peter the Great to the last days of the Romanoff dynasty).
Russian Ark is considered a first person filmic experience as the viewer identifies himself directly with the camera-figure.
3.2. Reasons for choosing Russian Ark
We have chosen Russian Ark (RA) as our case study in this paper because it is considered a revolutionary film on the technical level of
cinematographic production, since it was the first ever feature-length film to be filmed entirely in a single continuous shot (95 minutes)
without resorting to any montage14. The production of RA was made possible thanks to the high-definition digital steady-cam (with a
portable hard-drive capable of registering around 100 minutes of uncompressed footage) that was specially developed for this film. The
video images were later transferred to conventional 35mm film for cinema screen projection purposes.
Filming a feature-length film in a single-shot sequence constrained, or rather enabled Sokurov (the writer, director, visual concept and
principal image designer of RA) to elaborate a film grammar unique to him and this film; by omitting a well-established specific
cinematic code, Sokurov, in need of some sort of punctuation device to pace and structure his film, resorted to a non-specific code, i.e. the
Architectural code, and more specifically the architectonic codes of the Hermitage Museum. This means that the cinematic code of
montage in RA is replaced by the architectonic codes of spatial structuring. Thus the critical reason for our choice of RA is our hypothesis
that this film could play an intermediary role between film and architecture.
3.3. Segmentation of Russian Ark
Since RA is a single-shot sequence film, already “the smallest unit of the filmic chain”, we find that although we cannot segment it
directly following our elementary units of film model used to define “classic” films, we can at least use it as a reference.
Therefore, as a starting point we make the following hypothesis:
As defined earlier in Section 2.6, a single-shot sequence “is a continuous flow of images (and sounds) uninterrupted by editing”. It
means that the shot itself is made of smaller units. Therefore RA is made of audio-visual signifiers (images and sounds); and since it is a
narrative film, some of these signifiers must unconditionally convey narrative significance. In our case study, we consider audio-visual
signifiers to be narrative, those signifiers that denote characters and events that themselves denote specific historical time-periods. These
signifiers are also filmed with a camera in constant movement. Consequently, the correlations between these different forms of
significance (the audio-visual signifiers and camera movements) form syntagmatic relationships; and since in RA these narrative audio-
visual signifiers and camera movements are grouped together according to common themes (each group represents a specific character,
time-period and/or event), we consider these groups as narrative syntagmas (NS).
More specifically, we define as NS those syntagmas in which 1) the “Stranger” has meaningful conversations with other on-screen
characters, or 2) on-screen characters are recognized to have historical significance, or 3) sufficient meaningful intra-diegetic audio-visual
information is conveyed to the viewers in order to assimilate the represented time-period of the segment; thus the delimitation process of
the NS is done following the on-screen appearance and disappearance of these significant characters or the significance conveying audio-

− 344 −
visual elements. Consequently, all syntagmas that do not fulfill one of the above criteria are considered as non-narrative syntagmas
(NNS).
Table 1 Delimitation and Designation of the Narrative Syntagmas
In Table 1 we indicate each NS with a designation,
Narrative Time limits on the Referred
their position and limits on the time-track, the Syntagma Time-track Historical
Description of Syntagma
content
historical period they refer to, and finally a Designation (hh:mm:ss) Period
Peter the Great venting out his
description of their content. Then following the NSA 00:05:52-00:07:44 17th Century
anger
camera’s on-screen movements inside the Hermitage Young Catherine the Great
NSB 00:12:08-00:13:59 18th Century
admiring theater rehearsals
Winter Palace, we trace its actual trajectory on the NSC 00:17:02-00:21:35 21st Century Contemporary Figures
NSD 00:23:30-00:23:40 19th Century Brief appearance of Pushkin
floor plans of the museum in order to position the NS
Conversation with the Blind
in the corresponding spaces they appear in and also NSE 00:23:46-00:31:58
19th+21st Woman in the presence of
Century Russian Navy soldiers and
in the order of their appearance. museum clerks
Observing the distribution of the NS in the spaces Conversation with “the
NSF 00:34:44-00:38:30 21st Century
Talented Boy”
of the Hermitage in Fig. 3 we notice that in addition Conversation with ex-ballerina
NSG 00:40:24-00:42:36 21st Century
Alla Osipenko
to their clear limitations in actual screening time,
World War II segment 1:
they are also well defined in and limited to particular NSH 00:43:42-00:44:14 20th Century marching soldiers and overhead
airplanes
spaces. We therefore designate by E the group of all
World War II segment 2: the
spaces that contain NS, and by Ex a particular space NSI 00:46:46-00:48:14 20th Century
siege of Stalingrad
that contains NSy, and so we have Ex�NSy (x and y NSJ 00:49:36-00:52:10 18th Century Old Catherine the Great
indicate respectively the designation of a particular Nicholas Romanoff I receiving
NSK 00:54:20-01:00:31 19th Century
E and NS as specified in Fig. 3, e.g. E4�NSC); and we the Persian ambassador

designate by � the group of spaces that contain NNS. Discussion between 3

20th+ 21st
NSL 01:02:57-01:06:55 Generations of Hermitage
We also notice that the relation between Ex and Century
Directors
NSy is not always a “1 to 1” relation, as there are 6 Nicholas Romanoff II & Family
NSM 01:08:29-01:10:35 20th Century
cases where 1 NS occurs in 1 corresponding E, 5 having dinner
20th+ 21st
cases where 1 NS occurs in a corresponding n NSN 01:11:40-01:23:44
Century
Last Dance Ball in the Palace

number of E (where n>1), and where an m number of NSO 01:23:44-01:28:46 20th Century Final Procession

NS occur in 1 E (where m>1). Therefore we conclude

that in RA there are 3 types of syntagma-space relations:
� Type1��Ex�NSy (1 to 1 relations: 1 space includes 1 narrative syntagma)
� Type 2: nEx�NSy (n to 1 relations: a number of spaces include the same narrative syntagma)
� Type3: Ex�mNSy (1 to m relations: 1 space includes multiple syntagmas)
These types of syntagma-space relations in themselves constitute larger sets of syntagmas that we designate as narrative segments, or �
(where �z would be a particular � and z the designation of a � as indicated in Fig. 4) expressed in the following manner:
1. �z(T1): Ex�NSy (where � is a narrative segment composed of Type1 relations);
2. �z(T2): nEx�NSy (where � is a narrative segment composed of Type2 relations);
3. �z(T3): Ex�mNSy (where � is a narrative segment composed of Type3 relations).
Based on the typologies and expressions described above, we lay out the detailed compositions of �z in Table 2, and then place them in
space on the floor plans of the Hermitage Winter Palace in Fig. 4; noting that the on-screen temporal delimitations of �z are based on the
following 2 criteria:
1) Visual Limits (VL): start point is when Ex becomes on-screen even before being physically penetrated; end point is when Ex becomes
off-screen. These limits are considered as the limits of the �z.
2) Physical Limits (PL): start point is when the camera actually enters Ex; end point is when the camera actually exits Ex. These are
considered as the limits of the Ex.
We illustrate how these VL and PL manifest themselves in RA in the examples in Fig. 5.
Finally, having laid out in detail the temporal and spatial limits of the narrative segments �, what remains of the film as space and
time falls in the category of non-narrative segments (�) that are composed of non-narrative syntagmas. � are also positioned on the floor
plans of the Hermitage in Fig. 4.
We conclude this section by establishing the criteria (ref. Section 2.5), specific to RA, which delimit, identify and define its segments:
1) Unity of action: the on-screen, intra-diegetic manifestation of characters, events and sounds of narrative significance that represent a
same historical time period in spaces Ex;

− 345 −
 2) Type of Demarcation: The on-
screen limits of � (VL-PL) are
dependent on and in function of on-
screen, intra-diegetic architectonic
elements such as doors and thresholds;
3) Syntagmatic Structure: �z(T1; T2;
T3): nEx�mNSy (where �z is composed
of Type1 or Type2 or Type3 space-
syntagma relations, and where n�1
and m�1 but not simultaneously >1).
3.4. Analysis of Russian Ark’s Diegetic
Spatial Structure
As defined in Section 2.4 of this paper,
diegesis is the represented instance of
the narrative, the audio-visual
manifestation of its space and time
dimensions.
Just as an actual architectural
spatial structure is composed of the
total amount (addition) of all the
spaces that constitute it, so is the
Diegetic Spatial Structure (or Ds);
which means that the Ds of RA is the
accumulation of the group of spaces
that contain narrative segments (i.e.
E), and the group of spaces that
contain non-narrative segments (i.e. �),
expressed as: Ds=E+�. Consequently, Fig.3 Narrative and Non-narrative Syntagmas Positioning and Camera
Trajectory Original floor plans courtesy of Hermitagemuseum.org
Ds is any and all Hermitage space Designations added by the author
captured by the camera, visible on-
screen, and seen by the viewers (Fig. Table 2 Detailed Composition and Space/Time Delimitations of Narrative Segments
6).
Narrative Segment Visual Limits Physical Limits Composition of
At this stage, following the camera’s Designation (hh:mm:ss) (hh:mm:ss) the Narrative Segments
trajectory that we traced in Fig. 3, and
� �� 00:16:53 00:17:01 - 00:21:47 � �(T2):(E4+E5)�NSC
the result of placing the Ds on the floor � �(T2,T3):[(E7�( NSD+ NSE)),
� �� 00:23:09 - 00:31:13 00:23:17 - 00:31:01
plans, we begin to have a clearer view ((E7+E8+E9)�NSE)]
� �� 00:34:42 - 00:38:41 � �(T1):E13�NSSF
of the shape of RA’s diegesis’ spatial
� �� 00:38:56 00:39:26 - 00:45:02 � �(T3): E15�(NSG+NSH)
structure, which is linear. However
� �� 00:46:54 - 00:48:22 00:47:04 - 00:48:19 � �(T1): E17�NSI
this is made possible because of the � �� 00:49:03 - 00:52:36 � �(T2):(E18+E19)�NSJ
availability of Hermitage’s Winter � �� 00:54:00 - 01:10:03 00:54:20 - 01:00:43 � �(T1):E22�NSK

Palace floor plans, something that is � �� 01:03:03 01:03:07 - 01:07:07 � �(T1):E25�NSL

� �� 01:07:46 01:08:13 - 01:11:05 � �(T2):(E27+E28+E29)�NSM
not very common; which means that
� �� 01:11:06 01:11:51 - 01:23:58 � �(T1):E32�NSN
this method is not sufficiently general
� �� 01:23:28 01:23:58 - 01:29:15 � �(T2):(E33+E34+E35+E36)�NSO
to describe the architectonics of a
film’s diegetic structure.
We proceed then by following a relatively more generalized method of analysis, but keeping in mind the linear structure as a starting
hypothesis. Since all films are made of syntagmas, or at least a single shot, we use our findings concerning the segmentation and
syntagmas (Section 3.2) to demonstrate the above hypothesis. As previously discussed, the narrative and non-narrative segments of RA
are in designate as Diegetic Progression Graph, or DPG (Fig. 7a).

− 346 −
 The Abscissa (x axis) of the DPG
represents the actual function of the
screening time, and completely
dependent of, in correlation with, and
contained in the spaces of the
Hermitage; furthermore, the on-screen
(dis)appearance of the spaces being also
in function of the screening time of the
film, we place all of the above
interrelated elements (space-time-
segment) in a bi-axial function graph
that we screening time (T) of RA
(including opening and closing credits, a
total of 1hour 35minutes 21seconds), and
its units are hh:mm:ss. The values
placed on it are the Physical Time Limits
(PL) values from Table 1. The Ordinate
(y axis) represents the group of all
spaces filmed in the Winter Palace (Ds);
the numbers on it are not values, but the
designation of the spaces (Ex and �x) in
Fig. 3, and they consist of ordinate
segments and not points. As for the
origin O, it represents the Threshold
(Fig. 8, and E0 in Fig. 3) from which the
camera enters the Winter Palace that
occurs at t=00:14:17. And finally, the
bold lines with the Kana designations on Fig. 4 Narrative and Non-Narrative Segments Spatial Positioning
the left represent �z (narrative Original floor plans courtesy of Hermitagemuseum.org
Hatches and letter designations added by the author
segments) delimited in time and space,
and as for �z (non-narrative segments),
they are represented by the dotted lines
with the Kana designations on the right
(both designation based on Fig. 4). We
observe in the DPG that Ds is in
function of T, and �z and �z advance in
function of their corresponding
coordinates in time and space, giving us
the following: Ds=f(T), and �z(Ex;t) and
�z(�x;t).
Fig.5 Examples of VL-PL (Visual Limits-Physical Limits)
We give the following two examples of
�’s VL start/end when its corresponding E becomes on-screen/off-screen (Fig. 6, � � starts in
the actual space-time coordinates of a �z screenshot 2a when E7 becomes visible behind the threshold (left), and ends in screenshot
4b, when E8 is rendered off-screen behind the closed doors); as for its PL, the start/end is
for further clarification: marked by the camera’s physical entry/exit to/from its corresponding E (Fig. 6, screenshots
� �[(E4+ E5);(00:17:01,00:21:47)], and, 4a and 3b when the camera physically crosses the threshold).
Screenshots taken from Russian Ark
� �[(�6);(00:21:47, 00:23:17)].
Now, applying a Regression Model on the DPG (Fig. 7b), it appears that Ds=f(T) where f is a Linear Function. Since the value of T can
only be positive (before the origin O the film cannot exist), it is clear then that as time progresses positively (forward), �z and �z and their
corresponding spaces have to progress also forward. Using the model in Fig. 7b, we can also define the Spatial Velocity of RA as being the
angle �, where ��[0;90]. A larger value for � implies a larger number of Ds, � and � seen in a smaller amount of screening time T (and
vice versa).

− 347 −
 Based on the DPG we have demonstrated that the hypothesis proposed
earlier in this section was correct: the properties of RA’s diegetic spatial
structure are in fact linear, and progressive. Moreover, this structure is
expressed and regulated in relation to the � and � (spaces containing
narrative and non-narrative segments) put together.
3.5. Analysis of Russian Ark’s Narrative Organizational Structure (Syuzhet)
As explained earlier in Section 3.1 of this paper, the story of RA consists
most importantly of conversations, interactions with, and appearances of
characters and events referring with significance to different historical
time-periods (Table 1); and as discussed in Section 3.3, these characters
and events are considered in themselves the minimal narrative signifier,
i.e. narrative syntagmas (NS), contained in architectural spaces and
limited by architectonic elements. Moreover, as demonstrated in Section
3.4, the movement of the camera, and consequently the spatial structure of
the diegesis are linear and progressive, which means that the NS
themselves unfold linearly and progressively during the screening of the
film. Fig.6 Totality of spaces that contain narrative and non-
narrative syntagmas/segments: spaces captured by the
However, the time-periods represented in their corresponding NS are not camera, visible on-screen, and seen by the viewers (Ds).
arranged chronologically in RA, and a quick observation of Table 1 Original floor plans courtesy of Hermitagemuseum.org
Shading added by the author
confirms this: the NS in the table are classified in their order of
appearance (from top to bottom), and yet the historical periods they represent are non-chronological, and the number of narrative “time-
jumps” (the change from one historical period to another in the narrative of the film) seems to exceed the number of actual NS since three

Fig.7a DPG (Diegetic Progression Graph)
The progression of spaces (Ds), narrative and non-narrative
segments (� and �) is in function of the progression of the
time (T). The Kana designations of � and � are based on Fig.
4. The numbers on the x axis are designations of spaces
containing narrative and non-narrative syntagmas Ex and �x
(Fig. 3). The values of T are expressed in hh:mm:ss based on
the Physical Time Limits from Table 2.
Fig.7b�Regression Model applied on the DPG
Ds=f(T) is a linear function, and � represents the Spatial
Velocity Angle of Russian Ark. The Kana designations of the
� and � are based on Fig. 4. The numbers on the ordinate
axis are the designations of the spaces containing narrative
and non-narrative syntagmas Ex and �x (Fig. 3). The values
of T are expressed in hh:mm:ss based on the Physical Time
Limits from Table 2.

− 348 −
of them represent simultaneously 2 historical periods each (NSE, NSL and NSN).
In order to better understand these time-
jumps, specify their frequency and between
which historical periods they occur, and since
time-jumps are directional in theory, we use a
directional graph that we call Narrative Time
Fig. 8 Screenshots of the Threshold
Digraph 1 or NTD1 (Fig. 9). In NTD1, based Point of entry to the Winter Palace section our research is focused in. The door in the
on the classification in Table 1, we group dotted circle in screenshots 1 and 2 is the Threshold, which is E0 in Fig. 3, and the
origin O for the Ds axis in Fig. 7a and 7b. The man heading towards is the “Stranger”.
together all the NS that represent a similar Screenshot 4 is the point where the camera is considered to physically enter the
historical period. These groups are designated Winter Palace: this space is space E1 in Fig. 3, and Fig. 7a and 7b.
Screenshots taken from Russian Ark
as Gn, where n is the century the contents of G
represent (e.g. G19 is the group of NSy that represent the 19th Century), giving us the
following 5 groups: G17 (where G17�{NSA}), G18 (where G18�{NSB; NSJ}), G19 (where
G19�{NSD; NSE; NSK}), G20 (where G20�{NSH; NSI; NSL; NSM; NSN; NSO}), and G21
(where G21�{NSC; NSE; NSF; NSG; NSL; NSN}). And so Gn become the vertices of the
graph, while the arcs represent the time-jumps, designated as TJ.
Finally, the arcs are assigned numbers that represent the order of a time-jump’s
occurrence in RA.
In NTD1 we therefore have the following relation: TJi= (Gn,Gn’) where the arc TJi
is directed from Gn to Gn’ (i is the number assigned to the arc).
The non-chronological, non-linear nature of� the time-jumps that we observe in Fig.9�NTD1: Narrative Time Digraph 1
G17…G21: group of all the NS that represent the
Fig. 9 is further emphasized when represented in a second diagraph, Narrative same historical period (the number represents
Time Digraph 2 or NTD2 (Fig. 10), where the edges are weighted according to the the century). Numbers on arcs represent the
order of a time-jump’s occurrence.
number and direction of TJi that occur between two Gn (arcs in Fig. 9). NTD2 is also
organized based on the increase in the weight of the edges, from the smallest weight
(left) to the largest (right).
In this section we placed groups Gn of NS in the chronological order they would
have been in if they weren’t rearranged by the film’s syuzhet (or plot) (Fig. 9), while
the time-jump connections (the arcs in the graph) distorted their progressive
linearity; which means that this chronological organization represents the fabula (or
story), while the organization of the Gn connected by the time-jumps that gave us
NTD2, is one of the possible forms of the syuzhet, i.e. the structural organization of
the narrative, which arrange the NS and consequently the � in RA.�
�
Fig.10�NTD2: Narrative Time Digraph 2
4. Conclusions G17…G21: group of all the NS that represent the
same historical period (the number represents
� � In Section 3.3, we concluded that the syntagmatic structure of the narrative the century). Numbers on edges represent the
segments (the structure of the smallest significant elements in RA) is defined as number of directional jumps that occur between
2 groups.
�z(T1; T2; T3): nEx�mNSy, where �z is dependent on an actual architectural space Ex;
� � The demarcation of the segments is defined by the on-screen limits of � (VL-PL) that are set in function of on-screen, intra-diegetic,
actual architectonic elements such as doors and thresholds;
� � In Section 3.4, we found that the progressive linear form of the diegetic spatial structure of RA is expressed as a Linear Function of
Ds=f(T), which led us to conclude that �z and �z having the respective coordinates (Ex;t) and (�x;t), are dependent and in function of
actual architectural spaces;
� � And finally in Section 3.5, we found that the narrative spatial structure, or Syuzhet, the structure that gives form to, and rearranges
the story and all of its contents, is non-linear with digraph properties.
The first three points lead us to conclude that actual architectonic elements (from Hermitage’s Winter Palace), which are denoted
instances in the film, regulate, structure and give shape to the segments and diegesis of RA; this means that actual architectural spatial
units, depending on their arrangement, numbers and correlations could be considered similar to syntagmas and segments, while their
organizational spatial structure becomes similar to the diegetic spatial structure.

− 349 −
 As for the last point, it leads us to conclude that since narrative is the substantial dimension of narrative film, it could be similar to the
substantial dimension of architecture, which means the program (or primary function), and consequently the narrative organizational
structure (syuzhet) would become similar to the programmatic organizational structure of architecture.
From this point on, we will further deepen the findings of this paper concerning the conceptual similarities between segmentation,
diegetic and narrative structures of narrative film on one hand, and the architectonic spatial and programmatic structures on the other,
in order to develop a model of reference that can be used to understand and view films from an architectural perspective, and then
translate or express them in actual architectonic means.

Notes
1 Many architects often refer to cinema and film as they use windows and openings “framing” external views, or juxtaposing different functional or
qualitative spaces in linear series and calling them “sequences”. �
2 Studies approaching the field of cinema from a semio-linguistic angle are studies mostly comparing it to the spoken and written languages. Semio-
linguists and film theorists, such as, Christian Metz, Umberto Eco, Pier Paolo Pasolini, Gianfranco Bettetini, Roger Odin, Peter Wollen and Roland
Barthes, defined film and its contents, while determining its different significative components from the smallest units to the largest.
3 Metz, Christian: Film Language: A Semiotics of the Cinema, The University of Chicago Press, 1974
4 Denotation in film is the perceptual (audio-visual) similarity between the signifier and signified (Metz, Film Language: A Semiotics of the Cinema, pp.108).
5 Signifier according to De Saussure is the sensible, material, acoustic or visual signal which triggers a mental concept, the Signified. The perceptible aspect
of a sign is the Signifier; the absent mental representation evoked by it is the Signified, and the relationship between the two is Signification.
6 More specifically the original formulations of Victor Shklovsky (Eagle 1981)
7 “Diegesis” is translated into Japanese as: ��������������������
8 Metz, Christian: Film Language: A Semiotics of the Cinema, The University of Chicago Press, 1974, pp.98
9 Stam Robert, Burgoyne Robert, and Flitterman-Lewis Sandy: New Vocabularies in Film Semiotics, Routledge, 1992, pp.40
10 Metz, Christian: Film Language: A Semiotics of the Cinema, The University of Chicago Press, 1974, pp.106
11 Odin, Roger: Cinéma et Production de Sens, Armand Colin, 1990, pp.139
12 The actual time on the time-track on the Japanese version of the DVD registers 95minutes and 21seconds
13 Designation based on the name given to the character in the closing credits of Russian Ark
14 Alfred Hitchcock’s The Rope was the most notable attempt in producing a single-shot sequence film without editing/montage. However, technically it
cannot be considered as such, since Hitchcock used 3 film rolls to shoot the whole film (technical limitations of the celluloid films), and each roll of film is
connected to the other by means of editing techniques. Moreover the opening sequence contains one more editing point as the camera moves from an
external view to an internal one. Since these edits are hardly noticeable during the film, some analysts and theorists consider it as the first single-shot
sequence film.

References
1) Alexander, Lily: Storytelling in Time and Space: Studies in the Chronotope and Narrative Logic on Screen, Journal of Narrative Theory, Vol. 37,
No.1, pp. 27-64, Winter 2007
2) Aumont, Jacques, Marie Michel: L’Analyse des Films, Nathan, 1988
3) Deleuze, Gilles: Cinéma, Tome 1. L’Image-Mouvement, Broché, 1983
4) Deleuze, Gilles: Cinéma, Tome 2. L’Image-Temps, Broché, 1985
5) Eco, Umberto: Lector in Fabula, Bompiani, 1979
6) Eco, Umberto: The Role of the Reader: Explorations in the Semiotics of Texts, Indiana University Press, 1984
7) Eisenstein, Sergei M.: Montage and Architecture, Assemblage, No. 10, pp. 111-131, 1989
8) Gardies, André: L’Espace au Cinéma, Meridiens Klincksieck, 1993
9) Hedges, Ines: Form and Meaning in the French Film I: Time and Space, The French Review, Vol. LIV, No.1, pp.28-36, 1980
10) Jowett, Gareth, M. Linton, James: Movies as Mass Communication, Sage Publications, 1989
11) Knapp, Bettina L.: Archetype, Architecture, and the Writer, Indiana University Press, Bloomington, 1986
12) Manly Hopkins, Gerard, Pater, Walter: Literary Architecture: Essays Toward a Tradition, University of California Press, 1983
13) Martin, Wallace: Recent Theories of Narrative, Cornell University Press, 1986
14) Mele ,Francesco, Calabrese, Anotnio, and Marseglia, Roberta: Interactive Analysis of Time in Film Stories, AI*IA 2007, LNAI 4733, pp. 765-772, 2007
15) Metz, Christian: Film Language: A Semiotics of the Cinema, The University of Chicago Press, 1974
16) Neumann, Dietrich: Film Architecture: From Metropolis to Blade Runner, Prestel, 1999
17) Smitten, Jeffrey R., Daghistany, Ann: Spatial Form in Narrative, Cornell University Press, 1981
18) Odin, Roger: Cinéma et Production de Sens, Armand Colin, 1990
19) Sokurov, Alexander: Russian Ark, Hermitage Bridge Studio and Egoli Tossell Film AG production, 2002
20) Spottiswoode, Raymond: A Grammar of the Film, University of California Press, 1973
21) Stam, Robert, Burgoyne, Robert, and Flitterman-Lewis, Sandy: New Vocabularies in Film Semiotics, Routledge, 1992
22) Tudor, Deborah: The Eye of the Frog: Questions of Space in Films Using Digital Processes, Cinema Journal, Vol. 48, Number 1, pp. 90-110, Fall 2008
23) Vidler, Anthony: The Explosion of Space: Architecture and the Filmic Imaginary, Assemblage, No. 21, pp. 44-59, 1993
24) Winters, Ben: The Non-Diegetic Fallacy: Film, Music, and Narrative Space, Music & Letters, Vol. 91, No. 2, 2010
25) Wollen, Peter: Signs and Meaning in the Cinema, Indiana University Press, 1972
�
約
�����
�
� � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �
���� � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � ������������������������������
����������������������������� ������������������������������
������������������������������ ������������������������������

− 350 −
������������������������������
��������������������������������
� ������������������������������
��� ��������
���� ����
������������������������������
������������������������������
�������������������������������
���� ��������
�����������������������������
������������������������������
������������������������������
��������������������������������
���� ���
�����������������������������
������������� � ����������������
������������������������������
������������������������������
�����������������������
���� �����
�����������������������������
�������������������������
���� �������������������������������
�����������������������������
������������������������������
���������������������������� � �
�������������������
���� �������������
�����������������������������
������������������������������
������������������������������
������������������������������
�������������������������������
���� ����
�����������������������������
������������������������������
��������������������
� ������������������������������
�����������������
���� ����������������
�����������������������������
���� ������� �� �������������������
������������������������������
������������������������
�
�
�
�
�
�
���� ������������������
�����������������������������
������������������������������
���������������������
���� ����������������������
�����������������������������
������������������������������
������������������������������
������������������������������
������������������������������
�����������������������������
�NS�������� NS ������������������
�E������������������������(�)���
�������������������������������
�����������������������Ex�NSy , �z(T2):
nEx�NSy and �z(T3): Ex�mNSy����� NS �������(VL) �
������(PL)����
���� �������������������������
���������(Ds)������������������
������������������������������
�����������������������������
Ds=E+� �����������������������(DPG)�
����������������� Ds �����������
������������������������������
������������������������������
���������������������
���� �����������������������
�����������������������������
������������������������������
������������������������������
���������NTD1 � NTD2��������������
������������������������������
����������������������
�
�����
�����������������������������
������������������������������
������������������������������
�����������������������������
������������������������������
������������������������������
����������������������������
（2010年 7 月10日原稿受理，2010年11月15日採用決定）
− 351 −