Film style and narration in Rashomon

Abstract

This article analyses the use of film style in Rashomon (Kurosawa Akira, 1950) to determine whether the different accounts of the rape and murder provided by the bandit, the wife, the husband and the woodcutter are formally distinct by comparing shot length data and using multiple correspondence analysis (MCA) to look for relationships between shot scale, camera movement, camera angle and the use of point-of-view shots, reverse-angle cuts and axial cuts. The results show that the four accounts of the rape and the murder in Rashomon differ not only in their content but also in the way they are narrated. The editing pace varies so that although the action of the film is repeated the presentation of events to the viewer is different each time. Different types of shot are used to create the narrative perspectives of the bandit, the wife and the husband that marks them out as either active or passive narrators reflecting their level of narrative agency within the film, while the woodcutter's account exhibits both active and passive aspects to create an ambiguous mode of narration. Rashomon is a deliberately and precisely constructed artwork in which form and content work together to create an epistemological puzzle for the viewer.

Full text

Film style and narration in Rashomon
Nick Redfern
Department of Media, Film & Culture, Leeds Trinity University,
Brownberrie Lane, Horsforth, Leeds LS18 5HD, UK.
n.redfern {at} leedstrinity.ac.uk
https://leedstrinity.academia.edu/NickRedfern
Abstract
This article analyses the use of film style in Rashomon (Kurosawa Akira, 1950) to determine
whether the different accounts of the rape and murder provided by the bandit, the wife, the
husband and the woodcutter are formally distinct by comparing shot length data and using
multiple correspondence analysis (MCA) to look for relationships between shot scale,
camera movement, camera angle and the use of point-of-view (POV) shots, reverse-angle
(RA) cuts and axial cuts. The results show that the four accounts of the rape and the murder
in Rashomon differ not only in their content but also in the way they are narrated. The
editing pace varies so that although the action of the film is repeated the presentation of
events to the viewer is different each time. Different types of shot are used to create the
narrative perspectives of the bandit, the wife and the husband that marks them out as
either active or passive narrators reflecting their level of narrative agency within the film,
while the woodcutter’s account exhibits both active and passive aspects to create an
ambiguous mode of narration. Rashomon is a deliberately and precisely constructed
artwork in which form and content work together to create an epistemological puzzle for
the viewer.
Keywords: Kurosawa Akira, Rashomon, narration, film style, statistical analysis, multiple
correspondence analysis
This is a post-print of an article published as:
Film Style and Narration in Rashomon, Journal of Japanese and Korean Cinema 5 (1-2)
2013: 21-36. DOI: 10.1386/jjkc.5.1-2.21_1.
To cite this article please refer to the published version.

Film style and narration in Rashomon
[2]
Under the ruins of the Rashomon a woodcutter and a priest tell a commoner of a puzzling
case in which a woman was raped in the forest, her samurai-warrior husband killed, and a
bandit arrested for the crime. They recount four different versions of these crimes told by
the bandit, the wife and the husband (via a spiritual medium), and of the woodcutter
himself as an unseen witness to events. Each version of events differs from the others in
several key respects: the bandit claims that he killed the husband in a sword fight after the
wife demanded the two duel to save her from dishonour, a tale seemingly backed up by the
woodcutter who also describes a fight to the death at the request of the wife. The wife
confesses to murdering her husband with a dagger after he spurned her following the rape,
while the husband’s account concludes with his ritual suicide after the departure of the
bandit and the wife.
Since the film’s release critics have been fascinated by the irreconcilable narratives of
Kurosawa Akira’s Rashomon (1950) and have sought to define the meaning of a film that
permits no privileged interpretation. Surveys of contemporary reactions by American and
European critics by Smith (2002) and Van Es (2002) emphasize the film’s themes of
selfishness, the nature of truth, and the construction of social reality, often presenting
Rashomon to western audiences as an ‘Oriental’ enigma. Academic criticism has focussed
on the irreconcilability of the four narratives and the motives that drive the narrators. The
film has been understood as a statement on phenomenology that affords the viewer an
experience of the fact that nothing has a definite nature and that illusion is the way things
are (Linden 1973); a conundrum that strikes at the heart of our perception of human reality
(Castle 2003); and as a ‘vast distorting mirror or, better, a collection of prisms that reflect
and refract reality. … the world is an illusion, you yourself make reality, but this reality
undoes you if you submit to being limited by what you have made’ (Richie 1996: 76). The
source of the distortion in the different narratives is attributed to the egotism that drives
the participants to present the version of events that portrays themselves in the best
possible light, and though none can be said to be lying each shapes the facts to fit their
character and situation. Van Es draws these elements together, describing the film as ‘multi-
perspectivism with a moral edge’:
The film offers an interpretative labyrinth by showing us seven different stories from
six perspectives. On crucial points the stories are in conflict. This leads to an
unsolvable epistemological problem: we do not know what exactly happened in the
woods and who is responsible for what. In this sea of ambiguity people are thrown
back upon themselves. The two core themes of the film are the social construction of
reality and egoism (2002: 117).
However, Rashomon does not lead us to despair of the world. Yoshimoto (2000: 183–84)
interprets the film from a humanistic perspective by noting the source of the social chaos of
the film’s setting is the egotism that motivates each narrator to tell the version of events
that presents them in the best light, but which may be overcome by the capacity for human
compassion that leads the woodcutter to care for the foundling discovered in the film’s final
scene.
Rashomon’s view of human memory, perspective and narrative has been extraordinarily
influential, and the ‘Rashomon effect’ is a well-established psychological phenomenon of
interest routinely invoked as an explanation by anthropologists, sociologists and legal
researchers for differences in personal perspectives of a single event (see, e.g., Heider
1988). This effect has even been noted in statistical science when several models offer a
multitude of descriptions of a data set with about the same minimum error rates (Breiman
2001).

Film style and narration in Rashomon
[3]
Much less attention has been devoted to the functions of style in Rashomon and its role
in narration. Yoshimoto (2000: 185) writes that what makes Rashomon a special film is
‘first and foremost its formal experiment, particularly its audio-visual form and narration.
The focus of the film is how the story is presented as much as what it is about’. However,
there have been no attempts to discover if the four accounts of the rape and the murder are
stylistically distinctive. In part this is because style is considered by some critics to be of no
consequence to the film’s central premise about the nature of truth and reality, if it can be
said to be of note at all. Prince argues there is little to be discovered by looking at form in
Rashomon because it has no relevance to the central problem of the film:
The ambiguity within the film – the question of reliability of the various stories – is
psychological in nature, issuing from the characters and the reasons they have for
lying. It is not an ambiguity of form, located in the visual and aural organization of
the film. As Kurosawa said, the paradoxes of the film are those of the human heart.
They are not those of the image itself (1999: 131).
Similarly, Kovács (2007: 252) argues that in Rashomon ‘narration is considered problematic
not because of the problematic nature of narration but because of an extreme existential
situation where everybody lies’; and that the relativism of the film is moral in nature and
not something proper to art or narration. Some critics do not find much to admire in the
style of Rashomon. Japanese critic Tadashi Iijima thought the film to be a failure because of
‘its insufficient plan for visualizing the style of the original stories’ (quoted in Richie 1996:
80), while George Barbarow stated baldly that ‘what interests the director in this picture is
evidently the employment of whatever good idea he happens to think of, and these
improvisations are in without much regard for the entire pattern of the film. Indeed, the
picture’s pattern is merely fortuitous’ (1952: 421). Barbarow denies the film is a
masterpiece because a masterpiece ‘is not confused and confusing, as Rashomon is’.
A second factor is that when discussing film style in Rashomon in more positive terms
critics have subsumed variations in style between the different parts of the film under a
larger artistic scheme that maintains an aesthetic principle of formal unity: the idea of
‘threeness’. Tyler places Rashomon in a tradition of multi-perspectivism in modern art,
concluding the film’s aesthetic unity
lies in the fact that however different are the imaginations of the four witnesses,
whatever harsh vibrations their mutual contradictions set up, the general design (as
the film-makers have molded it) remains and dominates the work's final aspect of
great beauty and great truth (1987: 158).
Linden (1973) describes Rashomon as Kurosawa’s attempt to ‘exhaust’ the triangle, with
three different styles used for the three different parts of the film (a ‘conventional style with
many cuts and close-ups’ at the Rashomon, a ‘sedate compositional’ style for the court
scenes, and the use of a ‘fluid impressionistic montage’ in the forest sequences), along with
three different types of sound (Western music, Japanese music’ natural sounds). Yoshimoto
(2000: 185–88) also identifies a formal unity and coherence to the film based around the
number three as a structural element (three locations, three principal characters in each
location, the three days between the trial and the gate scenes, three characters in the gate’s
sign); and contrasting the vertical compositions of the gate scenes and the horizontality of
the trial sequences, while the scenes in the forest bring these two planes together with the
verticality of the trees set against the horizontal movement of the camera. Richie also
praises the film for achieving a formal unity despite being made up of ‘various parts, all of

Film style and narration in Rashomon
[4]
which work admirably together’ to produce ‘a kind of rhapsodic impressionism which from
time to time carries the story and creates the atmosphere’ and which is fully realized in the
woodcutter’s walk through the forest (1996: 77). He goes on to discuss the use of
contrasting shots, held for equal amounts of time, and of single close-ups to emphasize the
triangular nature of the story, while also observing that Kurosawa had probably never
moved the camera more than he does in Rashomon. The tendency to privilege formal unity
over formal variation as a measure of the quality of an artwork is common to discussion of
aesthetics in general and film style in particular, but it ill serves our understanding of a film
like Rashomon where the irreducibility of difference lies at the core of the work.
In this article I analyse film style in Rashomon to determine if the different versions of
events presented to the viewer are formally distinct by comparing shot length data and
using multiple correspondence analysis (MCA) to look for relationships between shot scale,
camera movement, camera angle and the use of point-of-view shots, reverse-angle cuts, and
axial cuts. Quantitative studies of literature have shown such methods to be effective in
distinguishing between different narrative voices within and between texts by examining a
large number of factors simultaneously (see, e.g., Tabata 1995; Stewart 2003; Hoover
2003), but this will be the first time such methods have been applied to the cinema. The
next section describes the variables used in the study and the statistical analyses employed;
and in the third section I discuss the editing structure of Rashomon, the relationships
between and functions of the different aspects of film style, and how style creates the
perspectives of the four narrators.
Methods
The version of Rashomon used in this study is that distributed on DVD by Optimum
Releasing (catalogue number: OPTD1131).
Variables
I collected data for one continuous variable (shot duration) and four categorical variables of
film style (shot scale, camera movement, camera angle and shot type) each with several
modalities. The duration of each shot was recorded in seconds, and, as a PAL DVD source is
used, I applied a correction factor of 1.0416 to the raw data. The twelve shots of the
Rashomon under the opening titles are not included.
I collected data on a range of categorical variables describing the properties of shots in
Rashomon. Scale of shot describes the distance of the viewer to the framed material, and I
use seven shot scales based on the relative position to the subject, which is typically the
human body (see Thompson and Bowen 2009: 12–20): big close-up (BCU), close-up (CU),
medium close-up (MCU), medium shot (MS), medium long shot (MLS), long shot (LS) and
very long shot (VLS). Camera angle describes the vertical position of the camera relative to
the framed material, with shots classed as either a LOW angle with the camera placed below
the subject looking up, NEUTRAL looking straight-on into a scene irrespective of camera
height, or a HIGH angle with the camera placed above the subject looking down onto a
scene. Rashomon uses several different types of camera movement but since the low
frequency of some movements may affect the statistical analysis I assign shots to one of four
broader categories: shots in which the camera does not move but the lens is rotated (RO) in
the horizontal and/or vertical planes (pan, tilt, pan-and-tilt, etc); mobile shots (MO) in
which the camera itself moves (tracking or dolly shots, etc); hybrid shots (HY) such as
track-and-pans in which the camera both moves and the lens RO; and static shots (ST), in
which there is no camera movement. Three shot types are included in the study. Point-of-
view (POV) refers to shots framed from the position of one of the characters such that

Film style and narration in Rashomon
[5]
viewer is able to see what he or she sees. Shots not framed from a character’s position in a
scene are classed as not point-of-view (X.POV). Reverse angle (RA) shots are photographed
from the opposite direction as the preceding shot, typically as part of a shot/reverse shot
pattern or POV shots. Shots that do not meet this definition are classed as not reverse-angle
shots (X.RA). Finally shots are classified as either being framed along or very close to the
axis of the lens (AXIAL) of the previous shot or not (X.AXIAL).
Not all aspects of style in Rashomon are amenable to this type of analysis. Mise-en-scène
is consistent across the film and does not vary from narrative to narrative, and while
performances may differ between tellings (e.g. Mifune Toshiro’s portrayal of the bandit in
the first and last narratives) they show no contrast at smaller scales (such as scene-by-
scene or shot-by-shot). Consequently, these aspects of film style cannot meaningfully be
described by a set of nominal variables (though it may be possible to do so for other films)
and so they do not form part of the analysis presented here.
Statistical analyses
Editing contributes to the experience of pace and rhythm in the cinema by controlling the
rate at which the viewer’s attention is directed, with changes in shot duration associated
with transitions between narrative sections, localized events of high dramatic import, and
the emotional tone and intensity of sequences (Dorai and Venkatesh 2001; Hanjalic and Xu
2005). (Motion magnitude and sound energy also contribute to the viewer’s experience of
pace in the cinema, but as yet there is no generally accepted method for quantifying these
features.) In this article, two methods are used to describe the editing of Rashomon. First, a
trendline is fitted to the time series using locally weighted regression (LOESS). LOESS
regression is a non-parametric method that uses a low-order polynomial to fit a trendline to
a time series by using only those data points in the neighbourhood of a specific point in the
time series rather than fitting a trendline globally. This neighbourhood is called the span
and is a fraction of the whole data set used to estimate the trend at a given point. The
resulting trendline provides a smoothed description of the shot lengths that reveals the
underlying structure from the noise of the raw data. Second, the five-number summary
describes the distribution of shot lengths in Rashomon as a whole and each of the
narratives, and by comparing these distributions we can determine if the duration of shots
in one part of the film tend to be shorter than the shots in another part. Specific sequences
in the narrative are described in terms of their number of shots (n), the total duration of the
sequence (Σ), the median shot length and the interquartile range (IQR). All times are given
in seconds (s).
Film style is a complex dynamic system comprised of many aspects (mise-en-scène,
cinematography, editing, sound, etc) and any proper understanding of how the different
elements of a film’s style are organized into its formal structure requires analysing several
variables simultaneously. This is a challenging prospect given the number of shots in a film
may reach the thousands while the number variables potentially relevant to any particular
analysis may also be very large. If each element of film form is considered in isolation the
full structure of the data collected about a film’s style will not be revealed, but applying
multivariate analysis allows us to overcome these problems so that we do not overlook
interesting features.
Multivariate statistical analysis is the simultaneous statistical analysis of a collection
of variables, which improves upon separate univariate analysis of each variable by
using information about the relationships between the variables. Analysis of each
variable is very likely to miss uncovering the key features of, and any interesting
‘patterns’ in, the multivariate data. (Everitt and Hothorn 2011: 2, original emphasis)

Film style and narration in Rashomon
[6]
Though such methods have not yet been applied to film style they have enormous potential
to expand our understanding of how form functions in the cinema.
To analyse the data collected on the multiple levels of the four categorical variables
described above I use MCA as a descriptive method of revealing patterns in complex data
sets by locating subjects and variables in low-dimensional space. MCA is an extension of
correspondence analysis which allows for the analysis of a range of categorical variables,
and is performed by applying correspondence analysis to the indicator matrix of n rows
(the individual shots) and m categories whose elements are 1 indicating the category of
each variable to which a shot belongs and 0 elsewhere (see Greenacre 2007: 137–44; Le
Roux and Rouanet 2010). Interpretation of the representation of the variables in MCA is
based on the geometry of points in low dimensional space: proximity between different
variables indicates they tend to occur together, and proximity between different levels of
the same variable indicates the associated groups of observations are similar. In this study,
MCA was applied to those shots that comprise the narratives of the rape and the killing as
told by the bandit, the wife, the husband and the woodcutter. Shots of the bandit, wife and
medium speaking to the judge in the courtyard are not included since these are at a
different level of narration and are not present in the woodcutter’s account. Not included in
this data set are the framing scenes under the Rashomon between the woodcutter, the
priest and the commoner, the woodcutter’s walk through the woods and his discovery of
the body, and the testimonies of the priest and the police agent. All the shots included in the
data set contribute to the construction of the axes. The narrators were added to the
variables map as supplementary points and so were not used in defining the distances
between the individual shots. MCA was performed using the FactoMineR package (Husson
et al. 2013) for R (R Development Core Team 2012).
Film style and narration in Rashomon
Figure 1 presents the time series for the whole film with the separate narratives indicated,
and Table 1 presents the descriptive statistics of the shot length data for Rashomon as a
whole and for each of the four narratives with the courtyard scenes removed. The main
feature evident in the time series of Rashomon is the shift to a slower cutting rate after the
end of the bandit’s testimony. Although he does not wish to ‘bore the reader with an
extended discussion of the number of seconds’ duration each phase of the film takes’,
Linden (1973: 400) claims to have noted a change in tempo over the course of the film and
that the ‘main blocks of the form move in accelerating time spans, thus adding momentum
to the movement of the whole’. However, it is clear from Figure 1 this claim is categorically
wrong. The editing over the course of the whole film actually slows down, and while the
tempo of the tales of the bandit and the wife do show a trend to shorter takes as they
progress this not the case for those of either the husband or the woodcutter which show a
rising trend in Figure 1. There is also a slight tendency for shots in the sequences under the
Rashomon in the latter half of the film to be longer than those in the opening scene of the
film: the 28 shots in the opening scene at the gate in which the commoner first encounters
the woodcutter and the priest have a median shot length of 11.8 seconds (IQR=11.3s), while
the 53 shots at the gate after the bandit’s tale tend to be slightly longer (median=14.5s,
IQR=17.2s). The longest takes in each of the sections associated with a narrator are those in
which the characters speak directly to camera in the courtyard scenes.

Film style and narration in Rashomon
[7]
Figure 1 Times series of Rashomon (1950). The four different narratives are indicated by
the shaded areas.
The shot length data (excluding the courtyard scenes) confirms the bandit’s version
tends to be edited more quickly than the other three versions and that the later sections of
the film are edited more slowly. The husband’s version shows much greater variation in
shot lengths than the other sections and, in particular, has more takes in the twenty to 40
seconds range. The four narratives can be split into two pairs based on the action
presented, with the bandit and the woodcutter narrating the sword fight and the wife and
husband narrating their emotional confrontation after the rape. The distribution of shot
lengths in each pair is opposed so that while each action is presented twice it features once
edited relatively quickly and then again more slowly. For example, the bandit’s account of
the sword fight is 25 shots long and runs from shot 188 to shot 212 for a total of 171.1
seconds, with a median shot length of 5.0s and an IQR of 6.9s; whereas the woodcutter’s
telling is edited much more slowly (shots 344–372, n=29, Σ=353.8s, median=9.6s,
IQR=10.5s). The sword fight as narrated by the woodcutter is altogether a more uncertain
affair than that recounted by the bandit, presenting the duellists as hesitant to initiate
combat and clumsy in their footwork and swordplay rather than the dynamism of the pure
physical action of the bandit’s account that flatters his vanity as conqueror of a samurai
warrior, and this difference is reflected in the pace of the editing. The difference in the
editing pace of the wife’s and husband’s narration also corresponds to the difference in the
emotional intensity of these sequences, contrasting the increasing cutting rate as the wife
directly confronts her husband with the passivity of the husband who must wait and listen
as his wife pleads with the bandit to accept her. This difference between the activity and the
passivity of narrators is a key organizing principle of the film and is discussed below.

Film style and narration in Rashomon
[8]
Table 1 Descriptive statistics of the complete shot length data for Rashomon (1950), and for
the four versions of events in the forest (data for the four narratives does not include shots
in the courtyard).
Rashomon
Bandit
Wife
Husband
Woodcutter
Shots
410
139
29
26
65
Length (s)
5375.4
1110.0
318.8
407.1
915.6
Minimum (s)
0.5
0.5
0.9
1.8
1.2
Lower quartile (s)
4.1
3.3
4.0
6.5
6.0
Median (s)
8.5
5.0
7.5
11.1
10.3
Upper quartile (s)
16.7
10.5
13.6
25.7
16.5
Maximum (s)
110.3
38.5
46.0
46.4
75.1
Table 2 presents the descriptive statistics for the categorical variables of film style in
Rashomon and for each of the four different narratives.
Table 2 Frequency of shot scale, camera movement, camera angle, and shot types for
Rashomon (1950), and for the four versions of events in the forest (data for the four
narratives does not include shots in the courtyard).
Rashomon
Bandit
Wife
Husband
Woodcutter
Shots
410
139
29
26
65
Big close-up
8
2
2
1
3
Close-up
60
20
8
4
12
Medium close-up
90
39
11
4
7
Medium shot
137
47
2
10
22
Medium long shot
17
4
1
1
5
Long shot
64
18
3
3
9
Very long shot
34
9
2
3
7
Rotated
64
28
0
2
12
Mobile
62
14
6
5
9
Hybrid
9
2
2
1
1
Static
275
95
21
18
43
Low angle
131
42
10
3
12
High angle
88
36
8
4
16
Neutral angle
191
61
11
19
37
Point-of-view
70
37
11
3
15
Reverse-angle
153
77
15
6
24
Axial
49
11
3
9
12
The map of the variables in Figure 2 shows there is a difference between shot types
associated with the first dimension (POV, RA, AXIAL) and the positioning of the camera
relative to the subject (SCALE and MOVEMENT) associated with the second dimension. This
is a distinction between the narrative function of shots and the visual qualities of those
shots, between those variables that describe the perspectival attributes of shots and those

Film style and narration in Rashomon
[9]
that describe presentational attributes. ANGLE is approximately equidistant from both
dimensions and has both perspectival and presentational aspects since angle is both a
matter of interpretation and of framing. Of course, scale and movement can also be used to
create perspective in narrative, but this does not appear to be the case in Rashomon. The
supplementary variable NARRATOR also lies close to the first dimension indicating that the
key factor in distinguishing between different narrators is associated with shot types used
to create the perspective of different characters rather than with presentational features
like scale and movement.
Figure 2 Variables map of film style for the four different accounts of the crimes in
Rashomon (1950) resulting from multiple correspondence analysis. ‘Narrator’ is shown as a
supplementary variable.
In order to get a better understanding of the relationship between film style and
narration in Rashomon it is necessary to look at the different categories of the variables.
Figure 3 presents the categories plot resulting from the MCA for the different accounts of
the crimes with the individual narrators added as supplementary points. The first
dimension contrasts the different shot types in the film. POV and RA shots are closely
related, a natural relationship since POV shots are typically framed from the reverse angle
to the preceding shot. Shots not cut along the lens axis are also closely associated with these
types of shots, while shots that do lie along the axis relative to the preceding shot are
associated with non-POV and non-RA shots. The narrators are orientated along the first
dimension according to how their perspective is created for the viewer. In terms of film
form, the bandit’s narration is most similar to that of the wife and both are most dissimilar
from the narration of the husband. The narratives told by the bandit and the wife feature a
large proportion of POV shots (27 and 38 %, respectively) and make extensive use of RA
cuts (55 and 52%), whereas shots framed along the lens axis relative to the previous shot
account for only 8 and 10%. In contrast, the husband’s narration uses fewer POV shots

Film style and narration in Rashomon
[10]
(12%) and RA shots (23%), and axial shots occur much more frequently (35%). The version
of events told by the woodcutter falls in between these extremes. In this final narrative, POV
(23%) shots and RA shots (37%) are more frequent than in that of the husband’s and occur
less frequently than in those of the wife and the bandit (albeit only slightly less in the case
of the latter). At the same time, axial cuts (18%) account for more shots than in the tales of
the bandit and the wife and for a smaller proportion of shots than in the husband’s tale.
Figure 3 Relationship between categories of film style for the four different accounts of the
crimes in Rashomon (1950). The narrators (the shaded squares) are shown as
supplementary points.
Camera angle follows the same pattern as shot types, with the participants of events in
the forest differing greatly from one another while the woodcutter exhibits features
common to all three. The bandit and the wife’s versions have similar profiles with a greater
proportion of high angle shots (26 and 28%, respectively) than in the husband’s sections
(15%), and they also have more low angle shots (30 and 34%) compared to the husband
(12%). The husband’s version is strongly associated with shots framed at a neutral angle
(73%) compared to the other two (44 and 38%). The woodcutter’s narrative exhibits a
stronger tendency to neutral shots (57%) than the bandit and the wife but at the same time
a weaker tendency than the husband. The proportion of high angle shots in the
woodcutter’s version (25%) is similar to that in both the bandit and the wife, but has fewer
low angle shots (18%).
The second dimension of Figure 3 contrasts shot scale and camera movement. There are
no particular relationships between these presentational variables and the different shot
types described above indicating that Kurosawa varies the positioning of the camera and its
movement when creating a character’s perspective rather than relying on a subset of
stylistic choices for each narrator. Nonetheless, there are some key differences in the use of
scale and movement between the different narratives.

Film style and narration in Rashomon
[11]
Static framing is opposed to those shots in which the lens is rotated and those in which
the camera moves, though hybrid shots are distinct from the other categories of camera
movement because such shots occur only occasionally in the sample (accounting for just 2
per cent) and the distance of this point from the origin and from the other movement
categories in Figure 3 is a reflection of their low frequency. Though the fluidity of
Kurosawa’s camera has been noted by many critics, the camera does not move in the
majority of the shots in the film and this is the case for all four narratives, in which the
camera is static for between 66 and 72 per cent of shots. However, when the frame does
move the type of movement differs between narratives. The narratives of the bandit and the
woodcutter have a greater proportion of shots in which the lens is rotated (which are
mostly pans), whereas those of the wife and the husband have greater proportions of
moving camera shots. This is due to the difference in the content of the narratives. In the
versions told by the bandit and the woodcutter there is considerable movement of
characters within the frame. In the bandit’s version, he and the husband run through the
forest to the glade where they fight for the first time, and he will also later drag the wife to
the same place to see her husband bound; and both the bandit and the woodcutter recount
the sword fight resulting in the husband’s death. The accounts of the husband and the wife
do not include the movement of the characters through the forest or the sword fight and
focus instead on the reaction of the husband to the rape of his wife. Instead of panning
shots, Kurosawa employs dolly shots to bring the viewer into the intimate space of the
couple. Camera movement in Rashomon is associated with what happens in the forest
rather than how those events are perceived by a particular character and functions purely
in presentational terms.
Looking at shot scales, close shots are opposed to distant framing in Figure 3. BCU and
medium long-shots rarely feature in any of the four narratives, and, in fact, account for very
few shots in Rashomon overall; while the proportion of LS and VLS are broadly consistent
across all four narratives and with the style of the film as a whole. The supplementary point
for the wife in Figure 3 lies slightly off the first dimension and shows some (small)
correlation with the second dimension thereby distinguishing her narration from those of
the male characters. This is due to a tendency for shots in the wife’s account to be framed
closer than in those of the other narrators. Unlike the versions related by the three male
characters where the most frequently occurring shot scale is the MS (between 34 and 38%),
in the wife’s tale only 7% of shots are framed at this scale; and the wife’s version has a
greater proportion of CU (28%) and medium CU (38%) than the others. This distinguishes
the wife’s version from that of her husband so that although they narrate the same events
the presentation of on-screen space is emotionally heightened in the wife’s narration while
the more distant framing in the husband’s tale physically distances him from his wife and
her attacker. There are no such notable differences between the narration of the bandit and
the woodcutter though they tell the same tale. Again this aspect of film style is not
associated with any particular shot type and functions presentationally.
The results of the statistical analysis of film style in Rashomon reveal the different ways
in which the perspectives of the narrators are created. The perspectives of the bandit and
the wife are created through the use of POV shots, providing direct access to a narrator’s
perspective so that the viewer sees what they see and, at the same time, is given access to
how they imagine others see them. For example, the wife’s testimony includes an exchange
of nine POV shots between her and her husband: the shots of the husband from the wife’s
position show the viewer what she saw (the contempt on her husband’s face); while the
shots framed from the husband’s position show us how the wife imagines herself to be seen
as she pleads for understanding. Similar exchanges occur early in the bandit’s narrative as
he first sets eyes on the wife and later in the three-way exchange of POV shots after the rape

Film style and narration in Rashomon
[12]
and between bandit and husband as they prepare to duel. This use of POV shots establishes
the bandit and the wife as active narrators, not merely recounting events of the past but
aligning the viewer physically and psychologically with their perspective.
Burch (1979: 298) notes that Kurosawa makes extensive use of contrasting shots and of
the 180°-reverse angle cut and it is clear that such editing plays an important role in
narration, but did not identify any variation in the use of this type of editing between the
different narratives. In the husband’s narrative it is the absence of such shots that stands
out. The husband is a passive narrator forced to watch events but unable to affect them, and
by using axial cuts in place of POV and RA shots Kurosawa shows us events happening
before the husband and then his response to them without admitting the viewer direct
access to his perspective. We do not see how the husband conceives of others or as he sees
himself in their eyes. POV editing depends on a two-shot structure showing both the glance
and the object-glanced-at (Carroll 1993), but Kurosawa’s use of axial cuts refuses to
complete this pattern either by showing the look on the husband’s face and cutting along
the axis to reveal both him and the object of his glance in a single frame divided into
background and foreground or by using over-the shoulder shots to conceal the husband’s
glance and then cutting along the axis to focus on the object of his attention. In both
instances we follow the husband’s glance but in neither case do we gain access to his
physical position in the grove because the RA cut is denied. Bordwell (2009) notes that the
axial cut is a ‘stylistic fingerprint’ of Kurosawa’s films, and it is clear that in Rashomon such
a spatial relationship between shots plays a specific function in creating a unique narration
for the husband. Kurosawa signals this difference in moving from the framing scenes at the
gate to the courtyard scenes and the narratives they contain: CUs of the woodcutter and the
priest are used to indicate which of these characters will relate to us whether they recount
the testimony of the bandit or the wife, but the disembodied narration of the deceased
husband (via the medium) is prefaced by two shots of the disarticulated head of a statue
linked by an cut along the axis of the lens.
This active/passive distinction is also evident in the use of camera angles: the use of
high and low angles in the bandit’s and the wife’s versions contribute to their narration by
defining relationships between characters in terms of their dominance and sub-ordinance
to one another, whereas the high frequency of neutral angles in the husband’s narration
refuses to establish these types of relationships for the viewer. This distinction also reflects
the level of agency of the characters within the film: the bandit initiates events in the forest
through his sexual desire for the wife while the duel for her honour is fought at the wife’s
insistence, but the husband has no such power and is forced to react to the other characters.
As a witness to events in the forest, the woodcutter is the source of the only first-hand
witness account available to the priest and the commoner (and, by extension, to the
viewer). His version of events is presented as an apparently objective account from a third-
person narrator standing outside events and able to describe the actions of those he
observes in all their complexity without the prejudice of their own self-serving
perspectives. The content of the story he tells combines some material from each the prior
three versions (the duel, the husband’s rejection of the wife, the wife’s flight, etc), thereby
corroborating some part of each without definitively ruling one version more reliable from
the others. At the same time his account employs elements of the narration from the other
three versions combining some of the active components of the testimonies of the bandit
and the wife and some of the passivity of the husband’s narration. POV shots are used in the
same way as the narration of the bandit and the wife, cutting between the bandit and the
husband as they stalk one another prior to the duel, or the fraught exchange of glances
between wife and husband; and axial cuts are used to show the character’s looking or being
looked at while refusing the glance/object structure of the POV shot, even repeating some

Film style and narration in Rashomon
[13]
of the set-ups from the husband’s section. One key difference from earlier narration is that
some axial cuts are now associated with the bandit as well as the husband, attributing to
him a more passive demeanour, especially at the beginning of the swordfight which (as
noted above) lacks the virility of his own narrative.
However, the truthfulness of the woodcutter’s narrative is in doubt. Because he has
already lied about his knowledge of events having originally claimed earlier in the film to
have discovered the body of the husband while walking in the forest and because he may
have stolen the dagger from the corpse of husband, we are not bound to accept the putative
objectivity of his account. This affects how we understand the narration in this part of
Rashomon, and, when considered in the context of the film as an ‘interpretative labyrinth’, it
becomes apparent that in this final version the use of film style in the narration of events
serves a different function than in the other three versions.
If we accept the thesis that egotism drives each character in their construction of the
narrative of events in the forest it is easy to see how the use of different shot types create
self-serving narratives for the bandit, the wife, and the husband that portray them in the
best light. The bandit lies to preserve his reputation as a courageous warrior and a virile
lover, the wife to preserve her virtue, the husband to maintain his social status by his
suicide. The narration of their accounts reflects the level of activity/passivity each lays
claim to in admitting to a crime to sustain their self-image. The woodcutter’s story is equally
self-serving but for very different reasons: he lies to conceal his own involvement and so his
tale may be nothing more than a mixture of three versions already told in an attempt to
draw attention away from his own culpability. By adopting elements of the narration from
the other versions he conceals his own position. Where the perspectives of the bandit, the
wife and the husband are known and conflicting, the perspective of the woodcutter is
unknown.
We do not know what events transpired between the bandit, the husband and his wife in
the forest. In part, this is because we have conflicting accounts from the three participants
that cannot be reconciled. But it is also because the woodcutter’s account does not help us
to solve this riddle, shrouding already uncertain events in another layer of uncertainty. The
woodcutter’s version is inherently ambiguous because it denies us the perspective we need
to make sense of the information in this and the other narratives.
Conclusion
In this article I used time series analysis and, for the first time in film studies, MCA to study
the functions of film style in Rashomon. The results show that Kurosawa varies the pace of
the editing so that although the action of the film is repeated the presentation of events to
the viewer is different each time; there is a distinction between presentational (shot scale
and camera movement) and perspectival (shot types) aspects of style depending on their
function within the film, while other elements (camera angle) fulfil both these functions;
and different types of shots are used to create the narrative perspectives of the bandit, the
wife and the husband that marks them out as either active or passive narrators reflecting
their level of narrative agency within the film while the woodcutter’s account exhibits both
active and passive aspects to create an ambiguous mode of narration.
A detailed analysis of film style and narration in Rashomon such as that presented here
does not solve the epistemological problem at the core of the film. Indeed, it is only by
examining simultaneously the various aspects of film style across hundreds of shots do we
begin to understand the fundamental role played by film style in creating the ambiguities at
the heart of the film through the representation of the perspectives of the different
narrators. Contrary to the opinion of Iijima, film style in Rashomon is carefully planned to

Film style and narration in Rashomon
[14]
present the subtleties of narrator’s perspectives of events in the forest. Contrary to the
opinions of Prince and Kovacs there is ambiguity at the level of the image and the relativism
of the film is a matter of perspective created by the material process of narration. The
pattern of the film is most certainly not ‘fortuitous’ as Barbarow claimed; rather it is a
deliberately and precisely constructed artwork in which form and content are unified in
creating the ‘Rashomon effect’.
References
Barbarow G 1952 Rashomon and the fifth witness, The Hudson Review 5 (2): 420-422.
Bordwell D 2009 Seed-beds of style, Observations on Film Art,
http://www.davidbordwell.net/blog/2009/11/27/seed-beds-of-style/, accessed 27
February 2013.
Breiman L 2001 Statistical modelling: the two cultures, Statistical Science 16 (3): 199–215.
Burch N 1979 To the Distant Observer: Form and Meaning in Japanese Cinema. Berkeley:
University of California Press, 1979.
Carroll N 1993 Toward a theory of point of view editing: communication, emotion, and the
movies, Poetics Today 14 (1): 123-141.
Castle R 2003 The radical capability of Rashomon, Film-Philosophy 7 (33):
http://www.film-philosophy.com/vol7-2003/n33castle, accessed 11 March 2013.
Dorai C and Venkatesh S 2001 Bridging the semantic gap in content management systems:
computational media aesthetics, in A Clarke, C Fencott, C Lindley, G Mitchell and F Nack
(eds.) Proceedings of the First International Conference on Computational Semiotics in
Games and New Media, 10–12 September 2001, Amsterdam, The Netherlands. Amsterdam:
CWI: 94–99.
Everitt B and Hothorn T 2011 An Introduction to Applied Multivariate Analysis with R. New
York: Springer.
Greenacre M 2007 Correspondence Analysis in Practice, second edition. Boca Raton, Fl:
Chapman & Hall/CRC.
Hanjalic A and Xu L-Q 2005 Affective video content and representation modelling, IEEE
Transactions on Multimedia 7 (1): 143–154.
Heider KG 1988 The Rashomon effect: when ethnographers disagree, American
Anthropologist 90 (1): 73–81.
Hoover DL 2003 Multivariate analysis and the study of style variation, Literary and
Linguistic Computing 18 (4): 341-360.
Husson F, Josse J, Le S, and Mazet J 2013 FactoMineR: multivariate exploratory data
analysis and data mining with R, R package version 1.23: http://CRAN.R-
project.org/package=FactoMineR, accessed 11 March 2013.
Kovács AB 2007 Screening Modernism: European Art Cinema, 1950-1980. Chicago: Chicago:
University Press.
Le Roux B and Rouanet H 2010 Multiple Correspondence Analysis. Thousand Oaks, CA:
Sage.
Linden GW 1973 Five views of Rashomon, Soundings: An Interdisciplinary Journal 56 (4):
393-411.
Prince S 1999 The Warrior’s Camera: The Cinema of Akira Kurosawa, second edition.
Princeton, NJ: Princeton University Press.
R Development Core Team 2012 R: A language and environment for statistical computing,
R Foundation for Statistical Computing, Vienna, Austria: http://www.R-project.org/,
accessed 11 March 2013.

Film style and narration in Rashomon
[15]
Richie D 1996 The Films of Akira Kurosawa, third edition. Berkeley: University of California
Press.
Smith GM 2002 Critical reception of Rashomon in the west, Asian Cinema 13 (2): 115-128.
Stewart LL 2003 Charles Brockden Brown: quantitative analysis and literary style, Literary
and Linguistic Computing 18 (2): 129-138.
Tabata T 1995 Narrative style and the frequencies of very common words: a corpus-based
approach to Dickens’s first person and third person narratives, English Corpus Studies 2:
91-109.
Thompson R and Bowen CJ 2009 Grammar of the Shot, second edition. Burlington, MA:
Focal Press.
Tyler P 1987 Rashomon as modern art, in D Richie (ed.) Rashomon. New Brunswick, NJ.
Rutgers University Press: 149-158.
Van Es R 2002 Persistent ambiguity and moral responsibility in Rashomon, in KL Stroehr
(ed.) Film and Knowledge: Essays on the Interpretation of Images and Ideas. Jefferson, NC:
MacFarland: 102-120.
Yashimoto M 2000 Kurosawa: Film Studies and Japanese Cinema. Durham, NC: Duke
University Press.